watervulnerability

AUTHORS AND CONTRIBUTORS 

Authors* 

Michael J. Furniss, Pacific Northwest Research Station 

Ken B. Roby, Lassen National Forest, retired 

Dan Cenderelli, Stream Systems Technology Center 

John Chatel, Sawtooth National Forest 

Caty F. Clifton, Umatilla National Forest 

Alan Clingenpeel, Quachita National Forest 

Polly E. Hays, Rocky Mountain Regional Office 

Dale Higgins, Chequamegon-Nicolet National Forest 

Ken Hodges, Chugach National Forest 

Carol Howe, Grand Mesa, Uncompahgre, and 

Gunnison National Forests 

Laura Jungst, Helena National Forest 

Joan Louie, Gallatin National Forest 

Christine Mai, Shasta-Trinity National Forest 

Ralph Martinez, Plumas National Forest 

Kerry Overton, Rocky Mountain Research Station 

Brian P. Staab, Pacific Northwest Region 

Rory Steinke, Coconino National Forest 

Mark Weinhold, White River National Forest 

* All are USDA Forest Service, with units specified 

Design and Layout 

April Kimmerly, Peters Kimmerly Design Associates 

Margaret Livingston 

The U.S. Department of Agriculture (USDA) prohibits discrimination 

in all its programs and activities on the basis of race, color, national 

origin, age, disability, and where applicable, sex, marital status, 

familial status, parental status, religion, sexual orientation, genetic 

information, political beliefs, reprisal, or because all or part of an 

individual’s income is derived from any public assistance program. 

(Not all prohibited bases apply to all programs.) Persons with 

Contributors and Reviewers** 

Christopher P. Carlson, Washington Office National Forest System 

Jim Morrison, Northern Regional Office 

Janine Rice, Rocky Mountain Research Station 

Jeremy Littel, Climate Impacts Group, University of Washington 

Dan Isaak, Rocky Mountain Research Station 

Charlie H. Luce, Rocky Mountain Research Station 

Linda Joyce, Rocky Mountain Research Station 

David Cleaves, Washington Office Research and Development 

Sherry Hazelhurst, Washington Office National Forest System 

Sarah J. Hines, Rocky Mountain Research Station 

John Potyondy, Stream Systems Technology Center 

Ryan Foote, Lassen National Forest 

Ann Rosecrance, Ventura River Watershed Council 

Derik Olson, Council on Environmental Quality 

** All are USDA Forest Service, with units specified, 

except Jeremy Littel. Ann Rosecrance, and Derik Olson 

Cover Photograph 

Recommended Citation 

North Pole Basin, Sopris Ranger District, White River National Forest, 

by Cindy Dean, White River National Forest 

Michael J. Furniss, Roby, Ken B., Cenderelli, Dan; Chatel, John; Clifton, Caty F.; 

Clingenpeel, Alan; Hays, Polly E.; Higgins, Dale; Hodges, Ken; Howe, Carol; Jungst, 

Laura; Louie, Joan; Mai, S Christine; Martinez, Ralph; Overton, Kerry; Staab, Brian P.; 

Steinke, Rory; Weinhold, Mark. 2012. Assessing the Vulnerability of Watersheds to 

Climate Change: Results of National Forest Watershed Vulnerability Pilot Assessments. 

Climate Change Resource Center. U.S. Department of Agriculture, Forest Service 305p. 

www.fs.fed.us/ccrc/wva 

Online Presentations by the pilot National Forests 

Two sets of oral presentations that describe the analyses were conducted and recorded: 

1) An early set of detailed presentations by each pilot National Forest, presented to 

other pilot Forest staff in Salt Lake City in September of 2010, can be found here: 

==> www.fsl.orst.edu/fs-pnw/wva/ Presentations are ~50 minutes each, and reflect 

analyses-in-progress. 

2) A series of 3 webinars were given in Januay and February of 2012 by each of the pilot 

National Forests. These are more advanced and refined than the previous set, but are 

shorter and have less detail. Basic recordings of the webinars may be found here: 

==> www.fs.fed.us/ccrc/livelearn/wva/ 

disabilities who require alternative means for communication of 

program information (Braille, large print, audiotape, etc.) should 

contact USDA’s TARGET Center at (202) 720-2600 (voice and TDD). 

To file a complaint of discrimination, write USDA, Director, Office 

of Civil Rights, 1400 Independence Avenue, SW, Washington, DC 

20250-9410 or call (800) 795-3272 (voice) or (202) 720-6382 (TDD). 

USDA is an equal opportunity provider and employer.

ASSESSSING THE VULNERABILITY OF WATERSHEDS 

TO CLIMATE CHANGE: 

Results of National Forest Watershed Vulnerability Pilot Assessments 

ABSTRACT 

Existing models and predictions project serious changes to 

worldwide hydrologic processes as a result of global climate 

change. Projections indicate that significant change may threaten 

National Forest System watersheds that are an important source of 

water used to support people, economies, and ecosystems. 

Wildland managers are expected to anticipate and respond to 

these threats, adjusting management priorities and actions. 

Because watersheds differ greatly in: (1) the values they support; 

(2) their exposure to climatic changes; and (3) their sensitivity to 

climatic changes, understanding these differences will help inform 

the setting of priorities and selection of management approaches. 

Drawing distinctions in climate change vulnerability among 

watersheds on a National Forest or Grassland allows more efficient 

and effective allocation of resources and better land and watershed 

stewardship. 

Eleven National Forests from throughout the United States, 

representing each of the nine Forest Service regions, conducted 

assessments of potential hydrologic change due to ongoing and 

expected climate warming. A pilot assessment approach was 

developed and implemented. Each National Forest identified 

water resources important in that area, assessed climate change 

exposure and watershed sensitivity, and evaluated the relative 

vulnerabilities of watersheds to climate change. The assessments 

provided management recommendations to anticipate and respond 

to projected climate-hydrologic changes. 

Completed assessments differed in level of detail, but all 

assessments identified priority areas and management actions 

to maintain or improve watershed resilience in response to a 

changing climate. The pilot efforts also identified key principles 

important to conducting future vulnerability assessments.

ASSESSSING THE VULNERABILITY OF WATERSHEDS 

TO CLIMATE CHANGE: 

Results of National Forest Watershed Vulnerability Pilot Assessments 

CONTENTS 

THE CHALLENGE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 

WATERSHED CONDITION, RESILIENCE, AND HEALTH . . . 

WHAT'S THE DIFFERENCE? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 

THE PILOT ASSESSMENT APPROACH . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 

IDENTIFY WATER RESOURCE VALUES AND SCALES OF ANALYSIS . . . . . . .4 

Water Resource Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 

Scale(s) of Analysis and Reporting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .6 

ASSESS EXPOSURE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 

Using Historic Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .7 

Climate Change Projections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .8 

Evaluating Hydrologic Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10 

Applying Exposure Projections . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .11 

EVALUATE WATERSHED SENSITIVITY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 

EVALUATE AND CATEGORIZE VULNERABILITY . . . . . . . . . . . . . . . . . . . . . . . 18 

IDENTIFY ADAPTIVE MANAGEMENT RESPONSES . . . . . . . . . . . . . . . . . . . .20 

CRITIQUE THE ASSESSMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23 

Data Gaps and Uncertainty . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23 

Other Lessons Learned . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .24 

SUMMARY . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25 

REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .26 

NATIONAL FOREST REPORTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .28 

Northern Region (R1): Gallatin National Forest . . . . . . . . . . . . . . . . . . . . .30 

Northern Region (R1): Helena National Forest . . . . . . . . . . . . . . . . . . . . . .46 

Rocky Mountain Region (R2): Grand Mesa, Uncompahgre, and 

Gunnison National Forests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .64 

Rocky Mountain Region (R2): White River National Forest . . . . . . . . . . .112 

Southwest Region (R3): Coconino National Forest . . . . . . . . . . . . . . . . . 130 

Intermountain Region (R4): Sawtooth National Forest . . . . . . . . . . . . . 158 

Pacific Southwest Region (R5): Shasta-Trinity National Forest . . . . . . . 185 

Pacific Northwest Region (R6): Umatilla National Forest . . . . . . . . . . . 210 

Southern Region (R8): Ouachita National Forest . . . . . . . . . . . . . . . . . . . 226 

Eastern Region (R9): Chequamegon-Nicolet National Forest . . . . . . . .236 

Alaska Region (R10): Chugach National Forest . . . . . . . . . . . . . . . . . . . .266

THE CHALLENGE 

Water and its availability and quality will be 

the main pressures on, and issues for, societies 

and the environment under climate change. 

—IPCC 2007 

Climate change poses important challenges to the 

US Forest Service (USFS), the agency charged with 

management of more than 193 million acres of public 

forests and grasslands. Current and projected trends 

in global warming present risks to a wide range of 

ecosystem values and services, and the impacts are 

most closely associated with water resources, including 

changes in volume, timing, and quality. 

In response, initial priorities of the USFS climate 

change strategy are to build knowledge, skills, and 

expertise, and to develop experience and partnerships. 

These initial steps build toward planning and designing 

management actions to improve ecosystem resilience 

(Furniss et al. 2010). In this report, and for the pilot 

analysis, the term “resilience” means both the resistance 

to adverse changes and the ability of a watershed to 

recover following adverse changes. 

Principles of Vulnerability Assessment 

Derived from WVA Pilots 

(detailed in boxes throughout report) 

1. Use resource values to focus the analysis. 

2. The HUC-6 is currently the best scale for 

analysis and reporting. 

3. Local climate data provides context. 

4. Analyze exposure before sensitivity. 

5. Don’t get lost in exposure data. 

6. Keep the end product in mind. 

Maintaining or improving resilience is widely accepted 

as the best means to adapt to climate change (Williams 

et al. 2007). Forest Service managers have extensive 

experience in implementing practices that improve 

watershed health and resilience, such as restoring 

connectivity to aquatic habitats, restoring degraded 

wetlands, and using prescribed burning to restore fire 

regimes. 

While much is known about the hydrologic impacts of 

climate change and the means to improve watershed 

resilience, linkages to integrate this knowledge 

into existing programs and priorities are needed. 

The capacity of National Forests and Grasslands 

1 | ASSESSING THE VULNERABILITY OF WATERSHEDS TO CLIMATE CHANGE 

Clean and abundant water is often considered the 

most valuable ecosystem service provided by the 

National Forests and Grasslands, and most climate 

changes affect hydrologic processes. Water from 

these lands is important for domestic, agricultural, 

and industrial uses, and for hydropower generation. 

It supports recreational uses and provides crucial 

habitat for threatened, endangered, and sensitive 

aquatic species. 

to implement effective management measures is 

constrained by available resources (budgets and 

staffing). Priorities that integrate the impacts of climate 

change are needed to effectively allocate resources and 

focus management activities. 

Climatic changes are not expressed uniformly across the 

landscape. Not all watersheds are equally vulnerable. 

Some support more water resource values and some are 

inherently more sensitive to change. Identifying these 

important differences is critical to setting priorities and 

identifying responses for management. 

Despite these challenges, Forest Service managers 

are being directed to act. The agency’s climate change 

strategy has been launched, and efforts to adapt to 

climate change are now a reporting requirement for 

Forest Supervisors. A Climate Change Scorecard 

measures progress made by each National Forest 

and Grassland in four areas, including assessment of 

resource vulnerabilities. 

Currently there are few examples of assessments that 

inform managers about vulnerability of watersheds 

to climate change. Existing assessments are limited 

to analyses of vulnerability of particular species or 

habitats (e.g., Gardali 2012). Likewise, existing protocols 

for vulnerability assessments (e.g., Glick et al. 2011) 

focus primarily on single species or specific biological 

communities. Informative examples of place-based 

assessments that provide relative ratings of vulnerability 

of watersheds to climate change are not available. 

In response to this information gap, the Forest 

Service Stream Systems Technology Center funded 

the Watershed Vulnerability Assessment (WVA) Pilot 

project to determine if watershed-focused climate change 

assessments could be prepared by National Forest staff, 

using existing data sources. The goal of the pilot project 

was to provide land managers with assessments of the 

relative vulnerability of watersheds to climate change. 

The project involved substantial collaboration between 

National Forest System and Research and Development

staff; the task group included representation from two 

Research Stations and each Forest Service Region. 

This report summarizes the pilot effort. Because each 

National Forest System unit has different levels of 

staffing and data availability, the results represent 

a diversity of approaches on how to conduct a 

vulnerability assessment. We provide an overview of 

core assessment components, and highlight similarities 

and differences of the eleven pilot assessments. We 

also share important concepts that emerged during 

completion of the pilot assessments. These “Assessment 

Principles” could be applied in assessments in other 

National Forests and Grasslands, and are described in 

boxes located throughout the report. 

Each individual pilot assessment is locally based and has 

relevance at local scales. We do not attempt to summarize 

all of the findings of these assessments; they are included 

as attachments to this report. The assessments represent 

a broad range of conditions similar to those found on 

National Forests and Grasslands across the country, 

and provide examples of approaches for a wide variety 

of environmental contexts. Readers are encouraged 

to review the individual pilot reports for details on 

methods used and results produced. 

WATERSHED CONDITION, 

HEALTH, AND RESILIENCE . . . 

WHAT'S THE DIFFERENCE? 

Everyone is entitled to their own opinions, but 

they are not entitled to their own facts. 

—Daniel Patrick Moynihan, US Senator (NY) 

For the purposes of this report, two frequently used 

terms—watershed condition and health—are considered 

interchangeable. Resilience is the capacity of a system 

to absorb disturbance and reorganize while undergoing 

change and still retain the same functions, structure, 

identity, and feedbacks (Walker et al. 2004). Because the 

term “resilience” is used most frequently in the climate 

change literature, we have used this term throughout 

this project and report. Watershed resilience can be 

described as a subset or synthesis of “watershed health” 

or “watershed condition” (Furniss et al. 2010). 

The Forest Service has recently published a methodology 

to assess watershed condition (defined in box below), 

and has conducted baseline assessments across the 

entire 193-million-acre National Forest System (USDA 


“Watershed condition is the state of the physical 

and biological characteristics and processes within 

a watershed that affect the soil and hydrologic 

functions supporting aquatic ecosystems.... When 

watersheds are functioning properly, they create 

and sustain functional terrestrial, riparian, aquatic, 

and wetland habitats that are capable of supporting 

diverse populations of native aquatic- and ripariandependent 

species. In general, the greater the 

departure from the natural pristine state, the more 

impaired the watershed condition is likely to be....” 

Watersheds that are functioning properly have five 

important characteristics (Williams et al. 1997): 

1. They provide for high biotic integrity, which 

includes habitats that support adaptive animal 

and plant communities that reflect natural 

processes. 

2. They are resilient and recover rapidly from 

natural and human disturbances. 

3. They exhibit a high degree of connectivity 

longitudinally along the stream, laterally across 

the floodplain and valley bottom, and vertically 

between surface and subsurface flows. 

4. They provide important ecosystem services, such 

as high quality water, the recharge of streams 

and aquifers, the maintenance of riparian 

communities, and the moderation of climate 

variability and change. 

5. They maintain long-term soil productivity. 

From the USFS Watershed Condition Framework 

(USDA 2011b) 

2011a). This national program was initiated concurrent 

with the pilot WVAs. 

THE PILOT ASSESSMENT APPROACH 

The scientist is not a person who gives the 

right answers; he’s the one who asks the right 

questions. — Claude Levi-Strauss 

The WVA pilot team was composed of watershed and 

aquatic specialists from each of the nine regions of the 

Forest Service, stationed on eleven National Forests 

(see Figure 1). The group was supported by a steering 

committee composed of representatives from two 

Research Stations and two Regional Offices. Pilot

Figure 1. Location of National Forests participating in the Watershed Vulnerability Pilot Assessments. Coordination 

was provided by representatives from Regional Office Staffs in Regions 2 and 6, and the Pacific Northwest and Rocky 

Mountain Research Stations. Note that the GMUG comprises the Grand Mesa, Uncomgahgre, and Gunnison National 

Forests. A parallel WVA was conducted on the Shoshone NF (Rice et al. 2012) and coordinated with the core WVA group 

of 11 pilots. 

National Forests were selected to provide a range of 

water resource issues and environmental factors, and 

each National Forest brought different levels of staffing, 

expertise, and existing information to the project. A 

few pilot Forests had taken initial steps to consider how 

climate change might impact management priorities, 

though most had not. The goal was to conduct pilot 

assessments with a range of analytical rigor, in different 

geographic settings and organizational structures, with 

varying subject-matter focus. 

The pilot team and steering committee met to develop a 

methodology to guide the assessments. The initial step 

was to define the purpose of the assessments, which was 

to identify (for each unit) areas with highest priority for 

implementing actions to maintain or improve watershed 

resilience. This approach is based on two assumptions. 

The first is that there is a strong correlation between the 

condition and resilience of watersheds, with watersheds 

in better condition displaying more resilience than 

comparable watersheds in poor condition. The second 


assumption is that climate change is one of many 

factors, both natural and anthropogenic, that affect 

hydrology and watershed condition. A conceptual 

model illustrating these factors and linkages is 

displayed in Figure 2. 

The objective of the pilot assessments stemmed from 

the need to prioritize where to concentrate management 

activities to improve or maintain resilience. Comparing 

analysis options against this objective helped National 

Forest staff focus their efforts. 

The process was intended to produce useful results 

with differing levels of data availability and resource 

investment. Given the variety of watershed types, water 

resource issues, experience, and data availability on the 

pilot Forests, a flexible assessment method was needed. 

The team developed an analysis method that relied 

heavily on previous experience with Watershed Analysis 

(USFS 1995) and the basic model of vulnerability (Figure 

2). The assessment steps are summarized in the box below.

Figure 2. Conceptual model for assessing vulnerability, showing linkages between exposure, values, and system 

condition (sensitivity). We found utility in separating 3 components of sensitivity. “Buffers” and “stressors” are humaninduced, 

while intrinsic sensitivity is based on inherent characteristics independent of human inf luence. 

Steps in the Watershed Vulnerability Assessments 

1. Identify water resource values and scales of 

analysis 

2. Assess exposure 

3. Evaluate watershed sensitivity 

4. Evaluate and categorize vulnerability 

5. Identify adaptive management responses 

6. Critique the assessment 

The pilot assessments benefited from having leaders 

identified at the outset of the project. Leaders 

coordinated the assessment on their units, and at times 

acted as a one-person analysis team. A defined project 

leader was important in making key decisions on what 

to include in the assessment, identifying available 

data, determining how to analyze the information, and 

making adjustments when necessary. 

Each pilot Forest took a slightly different approach, 

depending on the resources selected for analysis, the 

type and amount of data available, and the staff time 

that could be devoted to analysis. After the assessments 

were initiated, the pilot team met monthly via video 

conference to discuss progress and share ideas and 

approaches. These discussions led to changes in the 

stepwise process and to the methods used in individual 

assessments. 

For the WVA, vulnerability was defined as the interaction 

of climatic exposure with values at risk and watershed 

sensitivity. In the framework model, management actions 


are intended to increase the resilience or buffering capacity 

of watersheds by modifying the effect of stressors that 

decrease resilience. Each of the primary components of 

the assessment: values, exposure, sensitivity, vulnerability, 

and application and lessons learned from applying the 

conceptual model, are further described below. 

IDENTIFY WATER RESOURCE 

VALUES AND SCALES OF 

ANALYSIS 

Ecosystem management is most successful 

when it considers and connects all spatial and 

temporal scales. For collaborative analysis, 

a specific scale and unit of land must be 

chosen, but this does not imply that only the 

collaborative analysis scale matters: they all 

matter. —The authors 

Water Resource Values 

Identifying the water resources to be included is vital 

to the overall assessment. Water resources are the prism 

through which all the other assessment steps are viewed 

and focused. For example, factors used to characterize 

sensitivity and exposure are selected because they have 

strong linkages or they most directly affect the selected 

water resources. 

Each pilot Forest considered including assessment of 

at least three designated water resource values in their 

assessments. These were aquatic species, water uses 

(diversions and improvements), and infrastructure. The

Assessment Principle One: Use Resource Values 

to Focus the Analysis 

One of the major challenges in conducting a 

broad-scale analysis is deciding what to address. 

The land areas under consideration are large and 

ecosystems and social systems are extremely 

complex. Narrowing the focus of the pilot 

assessments was considered essential and was 

achieved by identifying key water resource issues 

using iterative analyses. 

One aspect of the approach instrumental in 

focussing the pilot efforts was using water 

resource values, identified at the outset, to drive 

the assessment. Once resources of concern are 

identified, assessment questions are narrowed. 

The question then is not what exposure attributes 

to use, but what exposure attributes have the 

strongest effect on the resource value. Likewise, 

the question “What elements influence watershed 

sensitivity?” narrows to “What watershed 

sensitivity elements most strongly influence the 

water resource?” 

Using a specific set of resource values as the 

prism through which exposure and sensitivity 

were evaluated also provided for comparison 

of responses between resource values. Often, 

analysts found commonality among resources 

and were able to combine resources and methods 

to streamline the assessment. 


rationale was that climate change would influence these 

resources in different ways, and that including them in 

the pilot analysis would broaden the range of analytical 

methods and approaches. 

Given this objective, each pilot Forest selected water 

resource values based on their importance and perceived 

susceptibility to climate changes. All pilot Forests 

included aquatic species (or habitat for selected aquatic 

species) and infrastructure in their analyses. Eight of the 

11 pilot Forests included the vulnerability of water uses in 

their assessments. The water resources addressed by each 

pilot Forest, and the reporting scale, are listed in Table 1. 

The species (and aquatic habitats) selected for analysis 

represent the range of aquatic habitats found on the 

pilot Forests. Anadromous fishes were a focus on each 

National Forest where they occurred. Other salmonids 

included in the analyses were red-band trout, bull trout, 

brook trout, and three species of cutthroat trout. Brook 

trout are of note: they were a resource of concern within 

their historic habitat on the Chequamegon-Nicolet 

NFs, and a stressor (invasive species) on several of the 

western pilot Forests. Other fishes included as resource 

issues were warm water species on the Ouachita and 

Coconino NFs. Amphibian species and habitat were 

included in three analyses. 

Region National Forest Scale of Analysis Reporting Scale Water Resource Issues 

1 Gallatin National Forest HUC-6 (subwatershed) Westslope cutthroat trout, 

Yellowstone cutthroat trout, water 

uses, infrastructure 

1 Helena National Forest HUC-6 (subwatershed) Westslope cutthroat trout, bull 

trout, recreational fisheries, 

infrastructure 

2 GMUG National Forest HUC-6 (subwatershed) aquatic habitats and species, water 


2 White River National Forest HUC-6 (subwatershed) boreal toad and cutthroat trout 

habitat, water uses, infrastructure 

3 Coconino Five HUC-5 

watersheds 

HUC-6 (subwatershed) amphibians, stream and riparian 

habitat, water uses, infrastructure 

4 Sawtooth Recreation Area HUC-6 (subwatershed) salmon, bull trout, water uses, 

infrastructure 

5 Shasta-Trinity National Forest HUC-6 (subwatershed) springs, salmon, redband trout, 

water uses, infrastructure 

6 Umatilla National Forest HUC-6 (subwatershed) springs, salmon, bull trout, water 


8 Ouachita National Forest HUC-6 (subwatershed) warm water fishes, infrastructure 

9 Chequamegon- 

Nicolet 

10 Chugach Eyak Lake and 

Resurrection Crk 

Watersheds 

National Forest HUC-6 (subwatershed) wetlands; cold, cool, and warmwater 

fishes; groundwater, infrastructure 

HUC-6 (subwatershed) salmon, hydropower, infrastructure 

Table 1. Water Resource issues, scope of analysis, and reporting scales included in pilot assessments

Figure 3. Density of springs and small lakes on the Shasta-Trinity NFs. Results are shown for HUC-4 (left), HUC-5 

(middle), and HUC-6 (right) scales. The Shasta-Trinity assessment evaluated resource value, sensitivity, and vulnerability 

at the three scales, all showing that identifying priority locations for management actions was best done by 

HUC-6. 

The evaluation of water resources resulted in maps 

and descriptions displaying the location and relative 

importance by subwatersheds for each resource or 

combination of resources. For example, the Shasta- 

Trinity NF analyzed the density of springs and small 

lakes at three watershed scales (Figure 3). The Sawtooth 

NF displayed the relative importance of infrastructure 

(road crossings and near-stream recreation facilities) by 

subwatershed in the Sawtooth National Recreation Area 

Figure 4. Amount of infrastructure (roads and developed 

recreation facilities) within the Sawtooth NRA. Redshaded 

subwatersheds have highest density of infra structure, 

yellow show moderate density, and green show the 

lowest density. Red lines are HUC-4 boundaries. 


Important Considerations in 

Assessing Water Resources 

• Identify partners who can improve the assessment 

and engage with them. 

• Identify the most important places (if possible), 

categorize their relative values (high, moderate, 

low), and map them. 

• Determine what relevant broad-scale evaluations, 

assessments, and plans are available. 

• Consider all downstream uses (such as species and 

diversions). 

• Identify any ecological thresholds or risk levels 

(flow requirements, temperatures, and so on) 

associated with specific resource values. 

• As the assessment progresses, look for similarities 

(and differences) in response of resource values 

and consider grouping resource values where 

appropriate. 

(Figure 4). The characterization of watersheds in terms 

of the resources they support is an important step in any 

watershed planning effort and a first step in informing 

managers where limited resources might be invested. 

The assessment goal was to identify the most important 

places, categorize their relative value (high, moderate, 

low), and map the individual and composite values. 

Scale(s) of Analysis and Reporting

The Pilot Assessments were conducted over relatively 

large geographic areas, typically (8 of the 11 pilots) an 

entire Forest. Three Forests analyzed smaller areas for 

specific reasons. The Chugach NF (R-10) focused on 

subwatersheds where management activities would be 

most influenced by the results of the assessment. The 

Assessment Principle Two: The HUC-6 is Currently 

the Best Scale for Analysis and Reporting 

The scale for the pilot assessments was not 

prescribed, but all pilot Forests elected to use 

the HUC-6 (subwatershed) scale to characterize 

and map results. Climatic exposure data was 

often available, displayed, and assessed at scales 

larger than HUC-6; the work by the Shasta-Trinity 

NFs demonstrated that HUC-4 and HUC-5 scales 

are usually too large to effectively manage for 

water values, sensitivity, adaptive capacity, and 

resilience. The HUC-6 is the appropriate size 

for planning and implementing management 

strategies to sustain or improve watershed 

condition. In addition, HUC-6 is also the scale 

used to assess and report conditions for the 

Classification portion of the Watershed Condition 

Framework. 

Coconino NF (R-3) included 5 watersheds (HUC-5) that 

support the majority of aquatic resources on the Forest. 

The Sawtooth NF limited its evaluation to the Sawtooth 

National Recreation Area because it supports remaining 

strongholds for steelhead, bull trout, and Chinook and 

sockeye salmon listed under the Endangered Species Act 

and significant data were available for this area of the 

forest. 

All of the pilot Forests used the subwatershed (HUC-6) 

scale for analysis and reporting. This arose from the 

shared conclusion that subwatersheds provide a logical 

unit and scale for setting priorities and implementing 

management activities on National Forest system lands. 

The Shasta-Trinity NFs also used the HUC-6 as the scale 

to apply results of the vulnerability assessment, and, in 

addition, evaluated water resource values, watershed 

sensitivity, and vulnerability at two broader scales 

(HUC-4 and HUC-5). The results from the Shasta- 

Trinity suggest that general trends can be expressed at 

broader scales, but as might be expected, detail shown 

at the HUC-6 scale is lost at each higher level (Figure 3). 


While HUC-6 was determined to be the best reporting 

unit for displaying water resource values, sensitivity, 

and vulnerability, exposure information is generally 

available and appropriately used only at broader scales. 

As a result, most pilot Forests evaluated exposure at the 

HUC-5 scale. 

ASSESS EXPOSURE 

So why worry about global warming, which 

is just one more scale of climate change? The 

problem is that global warming is essentially 

off the scale of normal in two ways: the rate at 

which this climate change is taking place, and 

how different the "new" climate is compared 

to what came before. —Anthony D. Barnosky 

The consideration of climate change exposure data is the 

primary difference between the WVA and evaluations 

that Forest Service professionals have previously 

produced. Past assessments have been conducted 

for watershed analysis, restoration planning, and 

watershed condition. Pilot team members built upon 

this experience but few team members had used or were 

familiar with climate change projections. 

Analysis of exposure included four components: 1) review 

and evaluation of pertinent local historic climatic data, 

Assessment Principle Three: Local Climate Data 

Provides Context 

Local or regional examples of historic changes 

in climate and to valued resources should be 

incorporated as components of the assessment. 

Such information (e.g., historic trends in 

temperature and precipitation, changes in ice 

duration, or species phenology) can readily 

illustrate current influences on water resources. 

These data are local, and usually of high confidence. 

Use of local and regional climatological data, field 

observations, and local knowledge helps to frame 

the importance of climate change in terms that are 

better understood and appreciated than relying 

only on model-based projections of future climate. 

2) selection and use of one or two modeled projections 

of future climate conditions, 3) analyses of historic and 

projected changes to hydrologic processes that might 

affect water resources, and 4) selection of metrics to 

analyze and display differences in exposure across each 

analysis area.

Using Historic Data 

One finding consistent to all the pilots was the value 

of local historic data in providing local context and 

understanding of climate change. Display of historic 

changes with strong connection to local water resource 

values is typically easier to understand and appreciate 

than projections of future conditions. Projections 

are uncertain becaues they are associated with future 

emission scenarios and modeling assumptions. 

Differences between models increase as they are 

projected multiple decades into the future and display 

high variability that may be unsettling to managers. 

Figure 5. Duration of Ice Cover (days) on Lake Mendota 

in Wisconsin, 1855-2008 

Snow Depth (in) Max, Mean & Min 

250 

200 

150 

100 

50 

0 

1945 

1955 

1965 

1975 

1985 

1995 

Klamath Province 

Trinity Basin 

West side of forest 

(22 stations) 

2005 

2015 


Historic data helps both analysts and decision-makers 

by providing local context and trends in climatic 

conditions. 

Two examples from the pilot assessments are included 

here. The first shows changes to ice cover on Lake 

Mendota in southern Wisconsin (Figure 5). This 

historical trend was obtained from the Wisconsin 

Initiative on Climate Change Impacts (WICCI) by 

the Chequamegon-Nicolet NF during the assessment 

process. The second example shows changes in snow 

depth from the Trinity and Sacramento River basins in 

the Shasta-Trinity NF (Figure 6). 

Climate Change Projections 

Evaluation of climate exposure was the most difficult 

component of the assessment for several pilot Forests, 

due primarily to lack of experience with downscaled 

global climate modeling data. There were two basic 

challenges: deciding which climate change projections 

to use, and selecting the climate metrics. 

The availability of downscaled climate model data has 

increased substantially since the WVA pilot project was 

initiated. Of particular note is data now available from 

the Climate Impacts Group (CIG) at the University of 

Washington. The CIG has evaluated available Global 

Circulation Models (GCMs), and determined which 

models and ensembles of models produce the best fit 

with historic data, for the major river basins of the 

western United States. Data provided by CIG were used 

Snow Depth (in) Max, Mean & Min 

250 

200 

150 

100 

50 

0 

1920 

1930 

1940 

1950 

1960 

1970 

1980 

1990 

2000 

2010 

Southern Cascades 

Sacramento Basin 

East side of forest 

(19 stations) 

Figure 6. Changes in average snow depths from snow courses located in the Trinity River basin (1945- 

2009) and Sacramento River basin (1930-2009). Blue is the mean, red is the minumum, and green is the 

maximum snow depth.)

for evaluations conducted in Regions 1, 2, 3, 4, and 6. The 

GMUG NFs used projections from CIG, and additional 

projections for the Upper Gunnison River (Barsugli 

and Mearns Draft 2010). In retrospect, providing data 

available from CIG at the outset would have expedited 

some analyses and greatly assisted the process. 

Increases in temperature (ºF) Percent change in precipitation 

B1 2050 B1 2080 A1B 2050 A1B 2080 B1 2050 B1 2080 A1B 2050 A1B 2080 

January 2.70 4.42 4.38 6.00 (0.69) 8.85 5.98 1.68 

February 3.50 4.01 4.46 5.19 (0.97) (4.50) (2.54) (1.24) 

March 3.46 4.25 4.70 5.74 (0.75) (4.30) 0.63 (5.17) 

April 2.99 4.46 4.49 5.93 5.42 2.45 (1.19) 0.67 

May 3.68 4.48 5.02 7.16 (8.46) (1.28) (6.26) (10.68) 

June 3.90 4.64 5.34 7.04 (5.87) (7.17) (8.76) (12.37) 

July 4.14 4.98 5.40 7.28 (8.34) (2.70) (7.39) (12.84) 

August 4.13 5.04 5.21 6.84 1.20 6.97 1.52 2.61 

September 4.23 5.49 5.35 7.45 (0.49) 1.10 (3.47) 1.32 

October 4.12 5.46 5.29 7.15 (13.81) (8.17) (9.75) (8.17) 

November 3.52 4.36 4.93 6.15 0.91 (5.08) (7.93) (8.75) 

December 3.18 4.40 4.11 5.97 5.20 (9.39) (1.69) (1.68) 

Annual 3.63 4.67 4.89 6.49 (2.22) (1.93) (3.40) (4.55) 

Table 2. Modeled exposure data from the Ouachita NF assessment showing monthly and annual changes in temperature 

and precipitation derived from the B1 and A1B climate scenarios and provided by The Nature Conservancy Climate 

Change Wizard. Decreases in precipitation are in parentheses. 

Legend 

2000 - 2009 

Resurrection Creek 

649 - 958 

958 - 1,099 

1,099 - 1,187 

1,187 - 1,308 

1,308 - 1,438 

Mean 985.28 

Legend 


2020 - 2029 


683 - 998 

998 - 1,099 

1,099 - 1,187 

1,187 - 1,308 

1,308 - 1,438 

Mean 1,017.5 

In Region 2, the White River NF used projections supplied 

by the Colorado Water Conservation Boards (Ray et al. 

2008, Spears et al. 2009). The Shasta-Trinity NFs (Region 

5) utilized the World Climate Research Programme's 

Coupled Model Intercomparison Project phase 3 

(CMIP3) multi-model dataset. This is a downscaled 

global temperature modeling output available from the 

University of California, Santa Barbara. The Ouachita 

Legend 

2050 - 2059 


775 - 998 

958 - 1,099 

1,099 - 1,187 

1,187 - 1,308 

1,308 - 1,438 

Mean 1,115.62 

Figure 7. An example of a downscaled, gridded projection of future climate conditions. The maps display projected 

precipitation (mm). The projections are based on the A1B model. The figure is from the Chugach NF WVA.

NF (Region 8) relied on information from The Nature 

Conservancy’s Climate Change Wizard (Table 2). The 

Chequamegon-Nicolet NF (Region 9) employed data 

from WICCI, and the Chugach NF (Region 10) utilized 

projections provided by the University of Alaska, 

Fairbanks (UAF) Scenarios Network for Alaska Planning 

Project (Figure 7). 

All the pilot assessments used air temperature change 

projections in their analyses and most pilots included 

projected changes to precipitation. These projections 

were obtained from the variety of publically available 

state or regional climate sources listed above. All 

projections of future climate are based on General 

Circulation Models (GCM). These models are 

mathematical representations of atmospheric and 

oceanic motion, physics, and chemistry, and employ 

different emission scenarios to yield predictions of 

temperature and precipitation change. The globalscale 

model outputs are very coarse, so data are 

often downscaled and used as inputs to macro-scale 

hydrologic models for use in regional and finer scale 

analysis, such as the WVA pilots. The accuracy of the 

data becomes more uncertain with each subsequent layer 

of modeling. The greatest certainty is associated with 

air temperature projections. Precipitation projections 

Projected Climatic Changes 

Anticipated Hydrologic 

Response 

Warmer air temperatures • Warmer water temperature in 

streams 

Changes in precipitation amounts 

and timing 

Less snowfall, earlier snowmelt, 

increased snowpack density 

Intensified storms, greater 

extremes of precipitation and wind 

• Altered timing and volume of 

runoff 

• Altered erosion rates 

• Higher winter flows 

• Lower summer flows 

• Earlier and smaller peak flows in 

spring 

• Greater likelihood of flooding 

• Increased erosion rates and 

sediment yields 


are highly variable, with even less certainty for derived 

attributes like snowmelt, runoff, and stream baseflows. 

Precise changes in hydrologic extremes, such as flood 

and drought frequency, cannot be credibly modeled at 

the stream reach scale at present. 

The WVA pilot experience points to the value of broader 

scale (e.g., Regional) vulnerability analyses in providing 

exposure data and recommending future climate 

scenarios to National Forests. Interpreting exposure 

data at a broad scale would be useful for several reasons. 

First, exposure data is not available at finer scales. 

Second, consistency among National Forests in selected 

emissions scenarios and modeling assumptions would 

allow comparisons of expected climate changes across 

National Forests. 

Evaluating Hydrologic Changes 

Using projections of future temperatures and other 

climatic changes, most pilot Forests then considered 

what specific hydrologic changes would result from 

projected climate changes, and how water resource 

values would be affected by these changes. This step 

was integrative. In addition to the obvious connection 

Potential Consequences to 

Watershed Resources 

• Decrease in coldwater aquatic 

habitats 

• Increases or decreases in 

availability of water supplies 

• Complex changes in water quality 

related to flow and sediment 

changes 

• Changes in the amounts, quality 

and distribution of aquatic and 

riparian habitats and biota 

• Changes in aquatic and riparian 

habitats 

• Increased damage to roads, 

campgrounds, and other facilities 

Table 3. Projected hydrologic changes relative to identified values (Helena NF). Adapted from Water, Climate change, 

and Forests GTR (Furniss et al. 2010).

etween exposure and water resources, the evaluation 

served to stimulate thinking about which watershed 

characteristics might influence (either moderate or 

exacerbate) the hydrologic response to climate change, 

providing a segue to evaluating watershed sensitivity. 

An example of this type of analysis, which tracks 

climate changes through hydrologic responses to 

potential impacts on water resources from the Helena 

NF, is provided in Table 3. 

Applying Exposure Projections 

Once the hydrologic processes important to selected 

water resource values were identified, exposure metrics 

closely linked to those processes were identified. 

The list of metrics used to characterize exposure was 

limited because of the commonality in water resources 

in the assessments (Table 4). The selection of exposure 

metrics differed among National Forests for three 

reasons. The first is the water resources themselves. 

The Chequamegon-Nicolet was the only pilot Forest to 

include assessment of changes to groundwater recharge, 

and the only pilot Forest to use soil-water balance as 

an exposure metric. The second difference in selecting 

metrics was data availability. Both the Sawtooth NF and 

Umatilla NF assessments included finer-scaled analysis 

of potential changes to water temperature in evaluating 

change to bull trout habitat. These analyses were possible 

because of the availability of stream temperature 

data and predictive models, and the support of the 

Rocky Mountain Research Station (RMRS). The third 

difference in exposure metrics was the level of analysis. 

Differences in the depth of analysis were partly the 

result of data availability, as in the example described 

above. But the amount of time team leads could devote 

to the assessment was also a factor. Available time and 

perceived need for detailed analysis were practical 

National Forest Exposure Metrics 

Gallatin Combined flow, Snowpack vulnerability 

Helena Winter water temps, Summer air temps, Snow Water Equivalent (SWE), Precipitation 

GMUG Seasonal temperature, Aridity index 

Ouachita Monthly precipitation and Temperature 

White River Snowpack vulnerability 

Sawtooth Winter peak flows, Summer stream temperature, Summer flows 

Coconino Snowpack vulnerability 

Shasta-Trinity Air temperature, Stream aspect, Snowpack vulnerability 

Umatilla Winter and summer temperatures, SWE 

Chequamegon-Nicolet Air temperature, Precipitation, Soil water balance, Rainstorm frequency and intensity 

Chugach Air temperature, Precipitation, Freeze and thaw days 

Table 4. National Forests and metrics included to evaluate exposure effects in Water Vulnerability Assessments 


Assessment Principle Four: Exposure Before 

Sensitivity 

The first iteration of the WVA process used in the 

pilot project called for assessment of sensitivity 

before evaluating exposure. The result was 

development of rather generic sensitivity factors 

that did not have the strongest links to hydrologic 

processes most likely to be affected by climate. 

Exposure is considered first in order to produce 

a list of the most important hydrologic changes 

affecting each water resource. Sensitivity 

elements that strongly modify these hydrologic 

changes are then selected. 

matters in the exposure analysis. Some pilots chose 

to make use of more detailed information that was 

available, and provided metrics with closer links to the 

subject water resources. Baseflow, for example, may be 

more closely linked with trout habitat than snowpack 

vulnerability but the decision to conduct more detailed 

analysis was largely driven by the anticipated need for 

adequate detail to rate watersheds and set priorities. 

Some analysts thought more detailed analysis would 

further discriminate areas at risk. Others thought the 

objective of rating watersheds could be met adequately 

with coarser evaluation of exposure. 

There are advantages and limitations to both the 

coarser and more detailed approaches to characterizing 

exposure. The common factor in cases where the most 

detailed analyses were conducted is that they evaluated 

effects on species which, because of population status 

and trend, were already the focal point of restoration 

strategies and management emphasis. In such cases, 

additional detail may be warranted. At the same time, 

exposure is the assessment component with the greatest

level of uncertainty. Though there may be uncertainty 

in characterizing resource value, especially when ratings 

comprise more than one resource (for example, frogs plus 

fish species), descriptions of resource locations generally 

have little error. Likewise, assessments of sensitivity, as 

we will see in the next section, are composites of both 

intrinsic and anthropogenic factors. Schemes to combine 

or weigh the factors contain error, relative to how these 

factors are expressed in nature. Nevertheless, these 

assessment components are likely to be more accurate 

than projections of future temperature and snowpack, 

especially regarding what will actually occur decades 

from now. 

On the White River, Gallatin, and Coconino NFs, 

changes to snowmelt hydrology were determined to 

be the primary hydrologic change affecting selected 

resources. In both the White River NF and Coconino 

NF assessments, changes to the existing snow line due 

to projected temperature increases were anticipated. The 

watershed area within zones of predicted snow elevation 

change was used to characterize relative exposure 

of subwatersheds. The Gallatin NF assessment used 

projected changes in snowpack from the CIG (Figure 8). 

The Gallatin NF also included assessments of changes to 

summer and winter flow (from VIC) in their assessment. 

A similar approach was taken on the Shasta-Trinity NF. 

The impact of climate change on stream temperatures 

and habitat for salmonids was the focus on four 

National Forests (Umatilla, Sawtooth, Helena, Shasta- 

Trinity). These pilots employed data that looked deeper 


at potential hydrologic changes than other assessments. 

The Chequamegon-Nicolet NFs included a salmonid 

(brook trout) in their assessment of potential impacts 

of stream temperature increases on 16 species of cold, 

cool, and warmwater fishes. In addition to stream 

temperature, the Sawtooth NF evaluated potential 

changes to frequency of flood flows critical to bull trout 

habitat condition. This evaluation was possible because 

of support from the Rocky Mountain Research Station, a 

leader in assessing potential climate change impacts on 

aquatic ecosystems. 

The Ouachita NF selected aquatic communities as the 

resource of concern, and identified increased sediment 

production as the most likely adverse effect to that 

resource. Changes in precipitation and temperature 

from The Nature Conservancy’s Climate Wizard (see 

Table 2) were captured by month from the composite 

climate change models. The predicted changes in climate 

were then used to modify the climate generator in the 

Watershed Erosion Prediction Project (WEPP) Model 

(Elliot et al. 1995), which were then used to estimate 

sediment production under different climatic scenarios. 

The Chequamegon-Nicolet NFs’ assessment considered 

how climate change might affect important aquatic 

habitats, including lakes and wetlands. A Soil Water 

Balance Model was used to assess how potential 

groundwater recharge might change in the future 

and whether any change will differ by soil type. A 

groundwater flow model will eventually be used to 

Historic 2040s 2080s 

Snowpack Vulnerability 

Rainfall Dominant 

Transitional 

Snowfall Dominant 

Change from Historic 

Snowmelt Dominant to Transitional 

Transitional to Rainfall Dominant 

Gallatin National Forest 

5th Level HUC 

Figure 8. Projected changes in snowpack vulnerability between historic and two future 

scenarios, Galatin NF. Data from CIG, using the CIG composite model

determine changes in groundwater levels and flow rates 

to lakes, streams, and wetlands. 

The analysis on the GMUG NFs differed from other 

assessments, in that results were displayed at a large scale. 

Six large geographic areas, stratified by climatic regime 

and elevation, were used for graphical analysis. Projected 

changes to maximum and minimum air temperatures 

and an index of aridity were factors used to rate exposure 

in each of these geographical areas. This analysis 

technique was at least partially driven by the resolution 

of the downscaled exposure data, which is typically on 

a grid of 6 km2 (Figure 7). This fairly gross resolution 

results in as few as two or three data points for a HUC-6, 

making discrimination at this scale inappropriate. As a 

result, pilots typically used HUC-6 for distinguishing 

differences in resource densities and sensitivity, overlaid 

with a larger-scale rating of exposure. 

Climate models typically provide predictions of 

temperature and precipitation. These data are then 

combined with characterizations of watershed 

characteristics and vegetation in modeling of 

other hydrologic variables. CIG has also developed 

predictions of hydrologic change based on the Variable 

Infiltration Capacity (VIC) model (Gao et al. in review). 

The CIG was extremely helpful in releasing data for 

use during the pilot study, and in explaining its utility 

and limitations. VIC is a distributed, largely physicallybased 

macro-scale model that balances water and energy 

fluxes at the land surface and takes into account soil 

moisture, infiltration, runoff, and baseflow processes 

within vegetation classes. It has been widely used in 


Assessment Principle Five: Don't Get Lost in 

Exposure Data 

Pilot Forests used exposure data of different 

specificity and detail (for example, in one 

case, only air temperature change; in another, 

predicted stream temperatures). The level of 

detail influenced the analysis, but the take-home 

message is that all levels of exposure projections 

produced useable vulnerability assessments. 

Detailed projections at management-relevant 

scales are not necessary to gauge relative 

vulnerability of watersheds. It is more productive 

to move forward with the analysis than to get lost 

in the details of refining exposure data. 

the western U.S. to study past and potential future 

changes to water flow regimes (e.g., Hamlet et al. 2009), 

snowpacks (Hamlet et al. 2005), and droughts (Luo and 

Wood 2007). Several pilots (Helena, GMUG, Coconino, 

and Sawtooth NFs) made use of the VIC model outputs 

to evaluate exposure. VIC attributes evaluated by pilots 

included runoff, baseflow, and snow water equivalent. 

Several pilots employed projections of changes to flow 

characteristics. These were selected because of their 

important influence on habitat for species of concern. 

Flow metrics were also useful in describing relative 

exposure of water uses. In contrast, predictions of peak- 

and low-flow responses to climate change are limited 

and consist primarily of generalized predictions of 

higher peaks and more severe droughts with warming 

Figure 9. Winter peak flow risk from Sawtooth NF WVA; current data (at left) and projected data for 2040 (at right). 

Ratings for subwatersheds are: highest risk (red), moderate risk (yellow), and lowest risk (green). Ratings were developed 

by assessing change to frequency of highest stream flows occurring during the winter. Red lines are HUC-4 boundaries.

climate (Casola et al. 2005). Only the Sawtooth NF 

applied a tool useful in describing exposure relative to 

increased peak flows and infrastructure. This analysis 

used the VIC-generated “Winter 95” metric. Winter 95 

represents the number of days during winter that are 

among the highest 5% of flows for the year. Winter was 

defined as Dec. 1 – Feb. 28. Changes in Winter 95 were 

determined by comparing the increase in the number 

of days with the highest 5% flows between current and 

predicted conditions (2040 and 2080). Subwatersheds 

with less than a 0.5-day increase were considered low 

risk, those with 0.5- to 2-day increases were considered 

moderate risk, and subwatersheds with increases greater 

than 2 days were considered high risk. Results of this 

analysis for 2040 are depicted in Figure 9. 

At first glance, the difference in metrics and the level of 

detail might suggest pilots took very different paths in 

their characterization of exposure. In fact, all the pilots 


Evaluating Exposure 

• Quantify trends in available, relevant, historical 

climate datasets. Local and regional data that 

display significant changes demonstrate the 

likelihood that changes will continue into the 

future. Observed changes and trends in ecosystem 

traits, such as ice duration, species phenology, 

and species composition, are of great value. 

• Use best available climate change projections. 

Focus on near-term time frames. 

• Identify the effects of a changing climate on 

watershed processes to inform iterations of 

exposure data acquisition and assessment. 

• Identify hydrologic processes important to the 

identified resource value. 

• Determine how projected changes in hydrologic 

processes might affect each resource value. 

• Quantify the relative magnitude of differences in 

effects, including spatial and temporal variability. 

• Include disturbance regimes in the analysis and 

quantify disturbance-related effects. 

• Document critical data gaps, rationale and 

assumptions for inferences, references for 

data sources, and confidence associated with 

assessment outputs. 


took very similar approaches. All used review of historic 

data to display the trend in local climatic conditions. All 

pilot Forests also first looked at projected temperature 

and precipitation changes. Sometimes this information 

had been compiled for states, sometimes for river basins 

or larger geographical areas. The commonality is that 

the analysis was broad-scale. Next, pilots considered 

what impacts the climatic changes would have on 

hydrologic processes, and then how the hydrologic 

changes would impact water resources. Differences 

in pilot outputs resulted from decisions made at this 

point, primarily influencing the level of detail used to 

characterize the hydrologic changes. Perhaps the most 

valuable lesson learned by pilots in assessing exposure 

was that you don’t have to become a climate scientist to 

do a climate change vulnerability assessment. 

EVALUATE WATERSHED 

SENSITIVITY 

Models are tools for thinkers, not crutches for 

the thoughtless. —Michael Soule 

The goal of assessing sensitivity was to place areas 

(subwatersheds) into categories based on how they 

would respond to the expected climate-induced changes 

to hydrologic processes. The sensitivity of watersheds 

to any change is partially a function of parent geology, 

soils, typical climate, topography, and vegetation. 

Human influences also affect watershed resilience, 

depending on the extent and location of managementrelated 

activities. 

The Forest Service often evaluates watershed condition; 

watershed specialists routinely describe watershed 

condition in NEPA analyses, and many National Forests 

have watershed or aquatic species restoration plans in 

place that weigh heavily on assessments of watershed 

condition. Several pilots were able to take advantage of 

existing watershed condition ratings and apply them to 

their WVA. This included use of the Blue Mountains 

Forest Plan revision watershed condition (Umatilla NF) 

and “Matrix of Pathways and Indicators” determination 

of watershed condition factors in conjunction with 

Endangered Species Act compliance for several 

salmonid species (Helena, Gallatin and Sawtooth NFs). 

Sensitivity indicators were selected that most influenced 

the hydrologic process and water resource value in 

question. Some indicators tend to dampen effects

Intrinsic Anthropogenic 

Geology Road Density 

Soil Types Road-Stream Proximity 

Risk of Mass Wasting Road Crossings 

Groundwater-Baseflow Range Condition 

Slope Water Diverted 

Aspect Vegetation Condition 

Table 5. Attributes most commonly used in assessing 

sensitivity by the pilot Forests 

(buffers) and others amplify effects (stressors). For 

example, road density may amplify peak flow response 

and the potential for flood damage to vulnerable 

infrastructure near streams. In contrast, investment 

in road improvements such as disconnection of road 

surfaces from streams would tend to buffer effects. 

Attributes selected by pilots to characterize sensitivity 

included both intrinsic factors and anthropogenic or 

management-related factors (Table 5). In some cases, 

pilots termed the “natural” factors as sensitivity, and 

the anthropogenic factors as risks, combining the two 

types of indicators to derive a measure of sensitivity. 

Most pilots included both types of indicators (intrinsic 

and anthropogenic), though two pilots (Chequamegon- 

Nicolet and Ouachita NFs) employed only intrinsic 

Uncompahgre 

Grand Mesa 

West Elk 

San Juans 

Legend 

Forest Boundary 

Geographic Areas 

Erosion Sensitivity Rankings 

Low 

Moderate 

High 

Upper Taylor 

Cochetopa 

Figure 10. Erosion Sensitivity Rating from the GMUG 

NFs WVA. This rating derived from subwatershed 

characterizations of runoff potential, rainfall intensity, 

stream density, density of response channels, and mass 

wasting potential. 


Figure 11. Influence of Hydrologic Soil Group (HSG) 

on groundwater Recharge (2046-2065 minus 1971-1990). 

HSG was one of two sensitivity attributes applied to the 

groundwa ter resource issue in the Chequamegon-Nicolet 

NFs W VA. 

factors, and the Coconino NF selected only factors 

that related to management activities. A sample output 

(relative erosion sensitivity of subwatersheds on the 

GMUG NFs) is shown in Figure 10. Soil hydrologic 

groups used to classify watershed sensitivity on the 

Chequamegon-Nicolet NFs are shown in Figure 11. 

The Chugach NF assessment used many of the same 

sensitivity attributes as the other pilot Forests, but 

the approach differed in that the analysis consisted of 

comparing two watersheds with significant management 

activity (primarily undeveloped watersheds or those in 

wilderness were not included). 

Ratings from National Watershed Condition 

Classification were used in the sensitivity evaluation by 

the Coconino and Gallatin NFs. The Coconino NF was 

completing the Condition Classification at the same time 

the WVA was underway, and staff realized the utility that 

data developed would have in both efforts. The Coconino 

NF used few intrinsic factors to characterize watershed 

Aquatic Biological 

Life Form Presence 

Native Species 

Exotic and or Invasive 

Species 

(Riparian) Vegetation 

Condition 

Terrestrial Physical 

Density 

Road Maintenance 

Proximity to Water 

Mass Wasting 

Soil Productivity 

Soil Erosion 

Soil Contamination 

Terrestrial Biological 

Fire Condition Class 

Wildfire Effects 

Loss of Forest Cover 

(Rangeland) Vegetation 

Condition 

(Riparian) Invasive 

Condition 

(Forest) Insects 

and Disease 

Figure 12. Watershed Condition Factors from the USFS 

Watershed Condition Classification. Aquatic-Physical 

attributes are not included. Factors shown in red are those 

used as sensitivity indicators in the Coconino NF WVA.

sensitivity. Most were derived from the Watershed 

Condition Classification (Figure 12). The Gallatin NF 

characterization of sensitivity had two components: 

one included intrinsic watershed attributes, the other 

included levels of disturbance. The Watershed Condition 

Classification ratings of “functioning,” “functioning at 

risk,” and “non-functioning” were used to characterize 

disturbance. Since National Forests and Grasslands now 

have completed the Watershed Condition Classification, 

this data would be useful in conducting future WVAs. 

Pilot Forests that took advantage of existing condition 

ratings tended to apply them to all resource issues. Several 

pilot Forests, however, identified different indicators for 

each resource value. While many indicators are important 

influences on multiple water resources, some are not. 

For instance, the most important factors affecting peak 

flows and infrastructure may differ from those that most 

influence springs and other aquatic habitats. 

Pilot Forests took several approaches to developing 

ratings of watershed sensitivity. In the simplest 

applications (Ouachita and Chequamegon-Nicolet 

NFs), sensitivity indicators (e.g., basin slope, peat land 

type) were used to place watersheds into different 

categories. Other pilot Forests produced sensitivity 

ratings based on numerous indicators. When multiple 

indicators were used, pilot Forests developed methods 

of weighting and rating the relative influence of the 

attributes. For example, when considering influences 

on stream habitat, the amount of water withdrawn 

from a subwatershed is likely more important than the 

condition of terrestrial vegetation, and would therefore 

be given greater weight in calculating a sensitivity score. 

One approach to weighting sensitivity indicators, from 

Subwatershed Attribute Type of Attribute Relative Weight 

Geochemistry of parent geology Inherent to watershed 0.25 Buffer 

Extent of glaciation Inherent to watershed 0.75 Buffer 

Aspect Inherent to watershed 0.50 Additive 

Hydroclimatic regime Inherent to watershed 1.0 Additive 

Weighted precipitation Inherent to watershed 1.0 Buffer 

Extent of surface water features Inherent to watershed 1.0 Buffer 

Extent of large-scale pine beetle 

mortality 


Sensitivity x Stressors 

Risk Ranking Matrix 

Stressors 

Net Effect Relative to 

Climate Change 

Inherent to watershed 0.5 Buffer (short term) 

Water uses Anthropogenic 1.0 Additive 

Development (primarily roads) Anthropogenic 0.5 Additive 

Extent of beetle salvage Anthropogenic 0.5 Additive (short term) 

Table 6. Summary of attribute types affecting subwatershed resilience to climate change (White River NF) 

Sensitivity 

Low 

Moderate 

High 

Low Moderate 

Low 

Low 

High 

High 

Low Low 

Low 

High 

High 

High 

Figure 13. Scheme used to rate watershed sensitivity 

on the GMUG NFs. The matrix combines ratings for 

water shed stressors and sensitivity. Ratings of erosion 

sensitivity (6 elements) and runoff sensitivity (7 elements) 

were combined to produce the sensitivity rating. Stressor 

rating was derived by combining ratings of past management 

(2 elements), roads (3 elements), vegetation treatments, 

private land, and mining. 

the White River NF assessment, is shown in Table 6. In 

several cases, pilot Forests distinguished intrinsic and 

anthropogenic factors, and used a categorical matrix 

Figure 14. Bayesian belief network for determining overall 

physical condition, from the Sawtooth NF WVA. Contri buting 

factors included habitat access, flow, channel condition, habitat 

elements, water quality, and watershed conditions.

approach to combining and categorizing sensitivity, into 

a single rating. An example of such an approach (from 

the GMUG NFs) is displayed in Figure 13. 

On the Sawtooth NF, assessment of watershed condition 

was aided by use of Bayesian belief networks (Lee and 

Rieman 1997). The networks were used to evaluate 

relative differences in predicted physical baseline 

outcomes. The basic structure employs a box-andarrow 

diagram depicting hypothesized causes, effects, 

and ecological interactions (see Figure 14). The system 

was used to weight the relative importance of, and 

connections between, a comprehensive list of intrinsic 

and management attributes, and watershed and habitat 

elements. As with the other pilot assessments, results 

from this process were used to rate watershed condition 

as high, moderate, or low. 

While there was no detailed comparison of the 

products developed from these varied approaches, 

they had one thing in common: they all made use of 

available information to the greatest degree possible. 

By using existing condition and/or sensitivity ratings 

or developing them from scratch, each pilot produced 

useful ratings of watershed vulnerability. It is very 

important to determine what intrinsic and managementrelated 

attributes are important influences on watershed 

condition as they strongly affect the selected water 

resources. If available data, analyses, or assessments 

include attributes that match those identified in the 

WVA process or may serve as surrogates, it makes sense 

to use them. 

Sensitivity 

Sensitivity 

High 

Moderate/High 

Moderate 

Moderate/Low 

Low 

Subwatersheds 

Umatilla National Forest 

Figure 15. Composite rating of watershed sensitivity 

from the Umatilla NF. Factors used in the rating include 

groundwater dependence, watershed restoration invest ment 

category, road density, near stream road length, road grade, 

range condition, forest vegetation condition, and aquatic 

habitat condition. White areas have no NFS ownership. 


Stream Habitat Sensitivity Rating 

Coconino National Forest WVA 

High 

Medium 

Low 

Figure 16. Watershed Sensitivity for Stream Habitats, 

Coconino NF. Attributes contributing to this rating were 

derived from the Forest’s Watershed Condition Classification. 

Stressors included water diversions, terrestrial 

vegetation condition, riparian vegetation condition and 

invasive species, road proximity to streams, and wildfire. 

Buffers included holding instream water rights and degree 

of implementation of regional groundwater policy. 


Evaluating Sensitivity 

• Determine the intrinsic factors (such as geology, 

soils, and topography) affecting the hydrologic 

processes of concern—those that can most affect 

the resource values. 

• Determine the management factors (such as roads 

and reservoirs) affecting the hydrologic processes 

of concern. 

• Determine if management activities will serve as 

buffers or stressors. 

• Consider weighing the relative importance of the 

buffers and stressors in influencing condition and 

response. 

• Evaluate trends or expected trends in stressors, 

and how management actions and restoration 

could affect them.

The sensitivity evaluation typically resulted in maps 

showing relative sensitivities of subwatersheds. Two 

examples of this type of product are displayed. Figure 15 

shows the sensitivity rating from the Umatilla NF, where 

(like the GMUG example) a matrix was used to produce 

a combined rating of intrinsic and anthropogenic 

factors. A combined sensitivity rating was applied 

to a composite of resource values. The Coconino NF 

developed different sensitivity ratings for each water 

resource issue (Figure 16). 

Recent trends and projected future trends in resource 

conditions should also be included. For example, 

increased water diversion could exacerbate effects on 

a resource, whereas anticipated road improvements 

could improve condition and reduce effects that might 

otherwise occur. 

EVALUATE AND CATEGORIZE 

VULNERABILITY 

Climate change is a risk-multiplier… any decline 

N 

NF Boundary 

HUC6 

WI Counties 

Low 

Moderate 

High 

Very High 

1:1,000,000 

Figure 10. Relative vulnerability of wetlands to climate change for HUC6 

watersheds on the Chequamegon-Nicolet National Forest. 

Figure 17. Classification of climate-change risk to wetlands 

on the Chequamegon-Nicolet NFs. The rating is based on 

the proportion of total wetland and acid wetland within the 

National Forest boundary in each HUC-6. Total wetland area 

ranged from 0 percent to 55.8 percent of the area for all HUC- 

6 watersheds. The HUC-6s with less than 10 percent were 

rated “low,” those with 10 percent to 30 percent were rated 

“moderate,” and those with greater than 30 percent were rated 

“high.” The HUC-6s with less than 5 percent acid wetland area 

were rated “low,” those with 5 percent to 15 percent were rated 

“moderate,” and those with greater than 15 percent were rated 

“high,” and above that value were "very high". These two risk 

classes were combined to form one vulnerability classification 

for each watershed. 


in the ecological resilience of one resource 

base or ecosystem increases the fragility of 

the whole —HRH Charles, The Prince of Wales, 

addressing UN climate conference COP15, 

Copenhagen (December 2009) 

A relative rating of vulnerability of water resources to 

climate change was produced by combining information 

from the evaluation of resource values, exposure, and 

sensitivity. Pilot Forests used a variety of approaches 

to complete this step. Primary determinants were the 

number of water resources selected for analysis, and 

the way values, sensitivities, and responses had been 

described. Some pilot Forests classified vulnerability 

based on a threshold or ecological value (such as the 

amount of wetland area in each watershed, as shown in the 

Chequamegon-Nicolet example in Figure 17). The most 

common approach used by pilot Forests was to merge the 

location of values with ratings of watershed sensitivity, 

and then overlay that summary rating with differences 

Composite Aquatic Resource 

Coconino National Forest WVA 

High Value, Sensitivity and Exposure 

High Value and Sensitivity, Moderate Exposure 

High Value, Moderate Sensitivity and High Exposure 

Figure 18. Vulnerability ratings from the Coconino NF. 

The map displays the subwatersheds with the highest 

density of water resource values (native fishes, amphi bians, 

water uses, stream habitat, riparian and spring habitat, and 

infrastructure) that also have high or moderate sensitivity 

and high or moderate exposure.

Geographic Areas 

W 

N 

0 5 10 20 Miles 

S 

Exposure 

Ranking 

E 

Value Risk Ranking 

(weighted Ave) 


Vulnerability 

Ranking* Adjusted Vulnerability Ranking** 

Uncompahgre 6 1 7/12=0.58 3 

Grand Mesa 5 2 7/12=0.58 4 

San Juans 4 6 10/12=0.83 6 

West Elk 3 3 6/12=0.50 2 

Upper Taylor 2 5 7/12=0.58 5 

Cochetopa 1 4 5/12=0.41 1 

* Exposure Ranking + Value Risk Ranking)/12 

** Upper Taylor adjusted > Grand Mesa and Uncompahgre (area in high risk) 

Grand Mesa adjusted > Uncompahgre (higher concentration of values) 

Table 7. Vulnerability ratings from the GMUG NFs. Exposure was ranked for the six landscape units (a composite of 

HUC-6 watersheds) on the Forest (1 is the lowest), based on biggest change in annual average maximum temperature, 

annual average minimum temperature, and percent change in annual aridity index. Value Risk Ranking is the highest risk 

to values based on weighted average of acres × count of high rankings for each subwatershed. 

Watershed Ranking 

Water Uses 

Legend 

Water Diversions 

A - Active structure with contemporary diversion records 

C - Conditional structure 

U - Active structures but diversion records are not maintained 

Water Uses Ranking 

High 

Moderate 

Low 

Watersheds Level 6 

Administrative Forest 

Figure 19. Climate change vulnerability rating for the water uses resource value, White River NF. Red shading depicts 

subwatersheds with the highest vulnerability. Points of diversion for water uses are shown as black dots.

in exposure. The result of combining these elements is a 

classification, typically by subwatershed (HUC-6), that 

displays relative vulnerability of the identified values. 

All pilot Forests provided a narrative and mapped their 

results. Some pilots combined resource values in the 

analysis (see Figure 18), and others displayed resource 

values separately (Figures 17 and 19). The GMUG NF’s 

summary rating of vulnerability was presented in tabular 

format (Table 7). The GMUG assessed exposure and 

rated vulnerability at the watershed scale. The GMUG’s 

adjusted vulnerability ranking combines the ratings of 

values, sensitivity, and exposure. 

Based on its strong partnership with the Rocky Mountain 

Research Station and its access to considerable habitat 

condition data (including stream temperature data), the 

Sawtooth NF conducted the most detailed evaluation. 

The Sawtooth NF analysis included assessing the 

effects of potential changes to stream temperature 

and flow on bull trout. Potential temperature effects 

were analyzed by summarizing the available stream 

miles that were within or exceeded 15ºC within each 

bull trout patch for 2008, 2040, and 2080 timeframes. 

Figure 20. Predicted bull trout persistence in 2040 for subwatersheds 

in the Sawtooth NRA. Red-shaded subwatersheds 

are at high extinction risk, yellow-shaded are at 

moderate risk, and green are at low risk (subwatersheds 

with out bull trout are not colored). Red lines are HUC-4 

boundaries. 


Impacts on both low flow and winter flows were also 

assessed. Change in mean summer flow was evaluated 

by looking at the percent change in flow from current 

to 2040 and 2080. Changes less than 20% baseflow 

were considered low risk, 20% to 40% were considered 

moderate, and greater than 40% were considered high 

risk. Winter flow analysis compared how the number of 

days with the highest 5% flows increased from current 

to 2040 and 2080. Subwatersheds with less than a 0.5day 

increase from current conditions were considered 

Assessment Principle Six: Keep the End Product in 

Mind 

A plethora of climate change exposure data is 

now available. Models are continually refined. The 

number, types, and detail of climate projections 

can be confusing and overwhelming. Managers 

and analysts should realize that projections have 

substantial uncertainty and that uncertainty 

grows with down-scaling and time. 

Most pilot Forests structured their analyses such 

that actual values for temperature changes, runoff 

changes, etc., were not critical. The focus was, 

instead, on the ranges and direction of projected 

changes. 

This approach was appropriate, because the 

objective was to produce a relative vulnerability 

rating to inform decisions about priority areas for 

management. 

Periodically reflecting on the goal—what decisions 

need to be informed—helps put the need for data 

and precision in perspective. The necessary depth 

of analysis is that which will produce these relative 

ratings. 

at low risk, 0.5- to 2-day increase were considered at 

moderate risk, and greater than 2-day increase from 

current conditions were considered at high risk. Once 

the individual elements were analyzed, a Bayesian Belief 

Network was used to rate the impact of the change on 

bull trout population persistence. The vulnerability 

rating resulting from this process (the extinction risk 

for bull trout) is shown in Figure 20. 

On the Ouachita NF, predicted changes to precipitation 

were applied to WEPP (Road) modeling of road sediment 

production and compared to modeled estimates of 

existing condition. Changes in future conditions were 

then compared to existing correlations between aquatic 

assemblages and sediment production, which include 

high, moderate, and low risk categories. In this case, the 

assessment illustrates the differences in risk categories

that exist presently, and those projected to result from 

climate change. 

As with some other aspects of the WVA, the Chugach NF 

took a different approach to assessing vulnerability, owing 

to the fact that they analyzed only the two subwatersheds 

in their analysis area that are subject to management 

activities. Their objective was to look at those two 

subwatersheds and identify specific vulnerabilities and 

possible mitigations. The Chugach assessment focused 

on potential changes to salmonid habitat. The creeks and 

tributaries in the two subwatersheds are currently cold 

enough that the projected increase in water temperatures 

would not exceed optimal temperatures. The key 

concern was the unknown response of other organisms 


Evaluating Vulnerability 

• Identify where the location of water resource 

values overlaps with highest sensitivity and 

greatest exposure. 

• Determine how changes in hydrologic processes 

affect water resource values. 

• Determine the relative vulnerabilities of 

watersheds across the assessment area (for 

example: low, moderate, high) to inform priorities 

for adaptive response to predicted climate change. 

(including aquatic invertebrates). Of specific interest was 

whether increased water temperatures would alter the 

life cycles of prey species currently synchronous with 

newly-emerged salmon fry. 

Pilot Forest staff brought a variety of skills and 

backgrounds to their assessments. Some assessments 

were prepared by teams, some primarily by one person. 

In addition, there was great variation in the types 

and amount of available information. Despite these 

differences, each pilot Forest was able to conduct an 

assessment and report the results in an effective way. 

There were many differences in how the individual 

steps were approached, and the results reflect these 

differences. All the vulnerability ratings were derived 

by combining values, sensitivity, and exposure. We 

believe that the performance of the pilots in completing 

meaningful assessments using the basic process should 

encourage other units desiring to conduct vulnerability 

assessments. 


IDENTIFY ADAPTIVE 

MANAGEMENT RESPONSES 

"Nobody made a greater mistake than he who 

did nothing because he could only do a little." 

—Edmund Burke 

Each pilot Forest produced an assessment that effectively 

displays the location of water resources of concern, key 

climate change metrics, and watershed sensitivities of 

the resources to the projected changes. The combination 

of these elements yielded relative ratings of watershed 

vulnerability to climate change. As such, the assessments 

met the objective of providing managers with information 

necessary to identify priority areas to undertake 

management actions. 

Management priorities should focus on maintaining 

or improving watershed resilience. Resilience is the 

capacity of an ecosystem to respond to a perturbation or 

disturbance by resisting damage and recovering quickly 

(Holling 1973). By definition, resilient watersheds are 

better able to continue delivery of ecosystem services 

when subjected to ecological change, including changes 

that might result from a warming climate. A related 

assumption is that watershed resilience is closely tied 

to watershed sensitivity. Watershed resilience is a 

product of both inherent sensitivity and anthropogenic 

influences on watershed condition. 

The results of the WVAs will be useful in development 

of management options and strategies. This includes 

discussion of which vulnerability classes should be 

STEP 5 

Monitor and 

Verification 

STEP 5 

Track Restoration 

Accomplishments 

STEP 1 

Classify Watershed 

Condition 

STEP 4 

Implement Integrated 

Projects 

STEP 2 

Prioritize Watersheds 

for Restoration 

STEP 3 

Develop Watershed 

Action Plans 

Figure 21. USFS Watershed Condition Framework. Watershed 

Vulnerability Assessments contribute directly to Steps 

1, 2, and 3.

highest priority for management actions. If, for instance, 

a highly-valued water resource has a very limited 

distribution, management options are limited. If the 

value is more widely dispersed, managers must decide 

if the most vulnerable areas should be highest priority, 

or if they should focus their efforts on sustaining the 

values in areas with lower vulnerability. Scale must be 

considered in this discussion. Naturally, for resources 

(especially species) whose range is greater than the 

analysis area, discussion of results with other land 

managers will be necessary. 

The greatest value of WVA results is in identifying 

geographical areas that are priorities for actions 

designed to maintain or improve watershed resilience. 

Several pilot Forests are already using the results to 

this end. The recently completed Watershed Condition 

Classification (Figure 21) led to the designation of 

priority subwatersheds for improvement actions. The 

connection of the WVA to setting these priorities is 

clear. One pilot Forest (Coconino NF) applied WVA 

findings during this priority-setting process. On many 

National Forests and Grasslands, strategic plans have 

been developed to guide restoration and management 

efforts. In these cases, the vulnerability assessment 

process will be used to reassess existing priorities and 

to determine if changes are warranted. None of the pilot 

Forests were engaged in Land Management Planning 

during the WVA, but results have clear application 

to that effort in helping to identify priority areas for 

management. 

One pilot Forest (Ouachita NF) incorporated potential 

management actions in their sensitivity ratings. A 

tabular result from the Ouachita NF (Table 8) displays 

the number of subwatersheds in different watershed 

risk classes. These ratings were derived from estimates 

of current and future sediment production. In this 

assessment, future changes to precipitation were 

predicted to increase sediment production, with 

subsequent impacts on aquatic communities. Shown are 

the projected change in watershed condition class caused 

Risk 

2010 

current 

2010 

with 

road 

mgmt 2050 B1 


by climate changes from conditions in 2010 to those 

projected for 2050 and 2080, and potential modifications 

to the response from road management activities. 

The analysis demonstrates that implementation of 

road management activities could reduce the number 

of subwatersheds with high risk ratings. In terms of 

setting priorities for management, subwatersheds where 

implementation of road improvements would reduce 

vulnerability should be considered for high priority. 

Road improvements were identified as a key action to 

improve condition and resilience of watersheds on all 

the pilot Forests. In addition to treatments that reduce 

erosion, road improvements can reduce the delivery 

of runoff from road segments to channels, prevent 

diversion of flow during large events, and restore 

aquatic habitat connectivity by providing for passage of 

aquatic organisms. 

As stated previously, watershed sensitivity is determined 

by both inherent and management-related factors. 

Managers have no control over the inherent factors, 

so to improve resilience, efforts must be directed at 

anthropogenic influences such as instream flows, roads, 

rangeland, and vegetation management. 

In the subwatersheds with the highest vulnerability, 

any activity that maintains or increases water quantity 

or quality would ultimately be beneficial. In addition 

to roadwork, management actions to maintain or 

improve resilience could include contesting new water 

rights proposals, exploring ways to convert existing 

water rights into instream flows, improving conditions 

in grazing allotments, restoring natural function in 

meadows, and implementing silvicultural treatments 

aimed at moving toward more natural fire regimes. 

WVA results can also help guide implementation 

of travel management planning by informing 

priority setting for decommissioning roads and road 

reconstruction/maintenance. As with the Ouachita NF 

example, disconnecting roads from the stream network 

is a key objective of such work. Similarly, WVA analysis 

Climate Change Scenarios 

2050 B1 

with 

road 

mgmt 2080 B1 

2080 B1 

with 

road 

mgmt 2050 A1B 

2050 A1B 

with 

road 

mgmt 

2080 

A1B 

2080 

A1B 

with 

road 

mgmt 

High 88 82 93 85 93 85 105 96 105 96 

Moderate 46 40 42 43 42 43 44 43 45 43 

Table 8. Vulnerability ratings (by risk class) of subwatersheds on the Ouachita NF, as influenced by climate change 

scenarios and application of road management actions (maintenance to standard and closure of user created trails)

could also help prioritize aquatic organism passage 

projects at road-stream crossings to allow migration by 

aquatic residents to suitable habitat as streamflow and 

temperatures change. 

Pilot Forests in the Rocky Mountains recognized the 

utility of WVA results in selecting the subset of highvulnerability 

watersheds in high pine-beetle-mortality 

areas. These areas are high priority for upgrades of 

road-stream crossings, to protect them from floods and 

debris flows. The same watersheds are also priorities for 

vegetation management to enhance natural reproduction, 

hydrologic recovery, stream shading, and future large 

woody debris recruitment. Both sets of actions would 

improve watershed condition and resilience. 

Not all the findings in the vulnerability assessments are 

good news. In some cases, projected changes may indicate 

that maintaining certain water resources (especially 

aquatic species) may be extremely difficult, even with 

restoration or improved management. In such cases, 

results from the vulnerability assessment may be used to 

rethink local or broad-scale improvement or protection 

strategies to prioritize limited management resources. 

The rating of vulnerability is based almost entirely on 

ecological considerations. Management activities are a 

key component in assessing watershed sensitivity, but 

only in terms of how they influence water resource values 

through hydrologic processes. While such characteristics 

are significant, social, economic, and administrative 

factors may be more important in determining where 

management activities can be effectively undertaken. 

Such factors include availability of expertise, land 

ownership, the presence of willing partners, LRMP 

guidance, and opportunities for internal or external 

funding. These factors need to be considered when 

determining where management activities should be 

focused. 

Almost all the pilot Forests encountered data gaps, and 

all encountered uncertainties during their analyses. Such 

data gaps (for instance, distribution of key species and 

uncertainty on road condition in key watersheds) can 

be used to identify inventory or monitoring priorities. 

Results from these efforts can, in turn, improve the 

utility of management and restoration plans. 

It is noteworthy that the specific management activities 

discussed above, including road improvements, 

improving aquatic organism passage, thinning forests 

to improve stand resilience, and improving range 

condition, are not new treatments designed to address 

climate change. Rather, they are activities where 

wildland resource managers have a long record of 

accomplishment. In land management and statutory 



Applying the Assessment Results 

• Consider whether additional information, analysis, 

or consultation is needed before setting priorities. 

• Identify approaches that can enhance resilience 

sufficiently to protect resource values. 

• Consider which effects of climate change might 

be irreversible, and how that can inform priority 

setting. 

• In places where vulnerabilities are high, can 

resource values be sustained? 

• Determine how management actions from WVA 

can be integrated into existing programs and 

priorities. 

• Identify management practices that would 

enhance resilience in both the short and long term, 

and assess the magnitude of treatment that would 

be required to meet improvement objectives. 

• Determine if land ownership patterns and 

administrative status of NFS lands are conducive 

to planning and implementing treatments. 

• Identify areas where partnerships would improve 

the likelihood of success. 

• Determine if sufficient technical and financial 

capacity is available to implement treatments. 

jargon, they are established Best Management Practices. 

Climate change increases the need for application of 

these practices nearly everywhere. 

CRITIQUE THE ASSESSMENT 

Test fast, fail fast, adjust fast. —Tom Peters 

The purpose of the WVA pilot was to determine if 

worthwhile assessments could be conducted with 

available information and expertise. Within a relatively 

short period of time and despite limited funding 

and other pressing business, watershed and aquatic 

specialists from the pilot Forests were able to develop a 

watershed vulnerability approach and complete useful 

assessments. Four pilot Forests were able to complete 

the process within 8 months, and an additional 5 were 

completed within a year. 

With an eye toward sharing approaches and 

experiences, each pilot Forest was asked to critique 

its assessment. These reviews are key in applying the 

principles of adaptive management to the vulnerability 

assessments. In this final section, we discuss how access

to information affected the assessments and share 

additional lessons learned by the pilots. 

Data Gaps and Uncertainty 

Assessing the sensitivity and vulnerability of watersheds 

to climate change is complex. At each step of the 

assessment process, pilot Forests encountered data gaps 

and uncertainty. Uncertainty was prevalent in estimates 

of exposure, but each analytical step contained some 

uncertainty (for instance, the expected response 

of hydrologic processes to climate change, and the 

response of aquatic resources to the hydrologic change). 

Lack of information also contributes to uncertainty; at 

the least, it limits the detail of the assessment. Every 

pilot Forest identified data needs, and each made 

assumptions about system responses and interactions. 

These were captured in the pilot reports as monitoring 

needs and included validating assumptions made in the 

assessment, tracking trends in key resource values, and 

providing data to inform key adaptive responses. 

Acquiring and applying data to improve the analysis 

would produce assessments with a higher level of 

confidence, but lack of data should not be used as a 

reason for not conducting an assessment. Some pilot 

Forests were data-rich, in terms of water resource, 

watershed sensitivity, and exposure information, and 

some were data-poor. In spite of these differences, 

all were able to apply the available information and 

complete a vulnerability assessment. Again, the 

objective of producing a relative rating of vulnerability 

(rather than a more rigorous, quantitative description of 

vulnerability) explains the favorable outcome. 

Other Lessons Learned 

The pilot assessment effort was successful in that it 

demonstrated that vulnerability assessments could 

be completed by National Forest staff using existing 

information and tools. Some of the primary reasons for 

this success have already been discussed; a few others 

are worth noting. 

While we have discussed how availability of data 

influenced the results, we have not articulated the 

importance of the format of the data. National Forests 

with digital data progressed much faster than those 

that had to convert paper summaries to digital formats. 

In some cases, lack of useable data caused elements of 

the assessment (for example, sensitivity factors) to be 

deferred. Ideally, necessary data would be gathered 

and prepared in anticipation of assessments. A credible 


The Forest Service Climate Change Resource Center 

The Climate Change Resource Center (www.fs.fed.us/ 

ccrc) provides land managers with an online portal 

to science-based information and tools concerning 

climate change and ecosystem management options. 

The CCRC’s objectives are to: 

(1) Synthesize scientific literature on ecosystem 

response, adaptation, and mitigation; 

(2) Highlight recent scientific research that has 

practical applications for practitioners on public and 

private lands; 

(3) Support communication of information through 

a user-friendly interface and appropriate use of 

multimedia; and 

(4) Work with scientists to develop educational 

resources. 

expectation that watershed vulnerability assessments 

will occur could help make this happen. 

Connection of the pilot Forests with ongoing climate 

change research and experienced scientists resulted in 

a more detailed analysis. The partnership between the 

Sawtooth NF and the Rocky Mountain Research Station 

yielded the most detailed pilot assessment. Collaborative 

work in downscaling climate change projections to 

assess potential changes in stream temperature was 

underway, and was well-utilized by the Sawtooth NF. 

Connection to sources of exposure data (especially CIG) 

also aided pilot Forests. 

As with most endeavors, the resulting products were 

strongly influenced by the experience and expertise of 

those participating. Those participants with the greatest 

localized knowledge of forest resources and interactions 

tended to have the easiest time with the process. Use of 

the pilot participants as trainers or facilitators for future 

assessments would streamline and focus those efforts. 

At the outset of the pilot project, several National 

Forests declined to participate, due to other priorities. 

The reality is that all participating National Forests 

managed to work on the WVA despite heavy workloads. 

Agreement to participate in the WVA process reflected 

recognition of the potential value of conducting 

WVAs. Pilot Forests where line and staff were more 

engaged with the assessments made resources (ID team 

members and GIS expertise) available to project leads, 

generally completed assessments sooner, and produced 

assessments of greater depth and detail. 

Recently-available electronic communication tools that 

facilitate information exchange proved extremely useful

to the pilot effort. The pilot project applied both a 

collaborative web space (a wiki bulletin board and a file 

repository) and videoconferencing to great advantage. 

Pilot leads were located across the country; monthly 

videoconferencing facilitated sharing of information 

and approaches, and helped cultivate a community 

of practice. The collaborative web space proved a 

very effective means of sharing written information, 

publications, announcements, and web links. The 

exchange of information enabled team members to learn 

from each other about processes and approaches that 

were working, and those that were posing difficulties. 

As a result, the individual pilot efforts were strongly 

influenced by each other. Readers who anticipate 

conducting an assessment are encouraged to contact 

members of the pilot assessment team, who can provide 

advice and counsel. 

Finally, it is clear that establishing an analytical 

methodology was of great value. As the WVA pilots 

evolved, participants made modifications to meet their 

needs, but the basic approach provided a consistent 

framework for pilots to apply. The success of the pilot 

Forests, which comprise a wide range of geographies, 

uses, and sensitivities, demonstrates that the conceptual 

basis of the approach is sound and likely applicable 

across the entire National Forest system. It will probably 

be applicable for all types of climate vulnerability 

assessments, not just water resources. Also of note is 

that we defined the component vulnerability terms 

at the outset, as this can be a source of confusion and 

unproductive debate; consistently sticking to and 

applying the terminology throughout the process 

assisted in moving the assessments forward. 

SUMMARY 

Observations clearly demonstrate that the Earth’s 

climate is warming and ecosystems are changing in 

response. Climate models predict substantial additional 

changes to world-wide temperatures and hydrologic 

processes throughout the 21st century. These changes 


will have complex and variable effects on the nation’s 

watersheds and impact their ability to sustain the 

ecosystem services upon which people depend. These 

effects pose significant challenges to the Forest Service 

and other land management agencies. To date, limited 

resources have been directed specifically towards 

planning for or responding to these changes on National 

Forests and Grasslands. This is largely due to the fact 

that managers have limited experience applying global 

or regional scale climate change information at the local 

scale. This has led to uncertainties about likely impacts 

and appropriate responses for individual National 

Forests or Grasslands. 

To address this need, we implemented a Pilot Watershed 

Vulnerability Assessment Project that developed and 

tested a process that National Forest personnel can 

use to complete useful, locally-based assessments of 

water resource vulnerability to climate change. These 

evaluations followed a process patterned after watershed 

analysis on federal lands in the Pacific Northwest (USFS, 

1995). The assessments covered relatively large areas (e.g., 

entire National Forests) with modest investments of time 

and effort. Regional climate projections, local historical 

data, and the Watershed Condition Classification recently 

completed by all National Forests in the US (www.fs.fed. 

us/publications/watershed), provided a solid base of 

information to support the assessments. 

In conducting the vulnerability assessments, forest staff 

became familiar with available historic climate data and 

climate projections for their geographic areas. Sorting 

through this information and learning how to use it 

was an important step in the process. Many forests 

found that partners had already compiled climate data 

and projections that could be utilized for the forest 

level assessments. Knowing future climates precisely or 

accurately is not possible, but this was not a barrier to 

producing effective, efficient, informative assessments. 

In addition to climate data, the pilot assessments used 

existing information on watershed sensitivity and water 

resource values, data that land managers are familiar

with and rely on in many resource decision making 

processes. The resulting assessments provided placebased 

identification of priority areas, with discernment 

of the watersheds most vulnerable and the most resilient 

to climate change. 

Assessing vulnerability is the essential first step in 

adapting to climate change, and this information 

provides a basis for managers to target investment 

of limited resources to sustain or improve watershed 

resilience. The good news is that the knowledge and 

tools to maintain and improve watershed resilience 

are already in place, while the National Watershed 

Condition Framework (USDA, 2011a) serves as 

a foundation for setting priorities and restoring 

watersheds and watershed services. Other US Forest 

Service programs to improve watersheds, meadows, and 

streams include diverse partners and programs across 

the country (Furniss et al. 2010). Implementation of this 

wide array of management activities is supported by 

decades of technical experience in planning, analysis, 

and collaboration. These existing core strengths can be 

effectively applied to address the growing challenge to 

public natural resources posed by our changing climate. 

REFERENCES 

Barsugli, J.J. and L.O. Mearns. Draft 2010. Climate and 

Hydrologic Change Scenarios for the Upper Gunnison River, 

Colorado. Prepared for The Nature Conservancy in support 

of the southwest Climate Change Initiative’s Climate Change 

Adaptation Workshop for Natural Resource Managers in the 

Gunnison Basin. 

Casola, J.H.; Kay, J.E.; Snover, A.K.; Norheim, R.A.; 

Whitely Binder, L.C.; Climate Impacts Group. 2005. 

Climate impacts on Washington’s hydropower, water supply, 

forests, fish and agriculture. Seattle, WA: Centre for Science 

and the Earth System, University of Washington. 43 p. 

Elliot, W.J., R.B. Foltz, and C.H. Luce, 1995. Validation 

of Water Erosion Prediction Project (WEPP) Model for 

Low-Volume Forest Roads. Conference Proceedings 

Sixth International Conference on Low-Volume Roads, 

Minneapolis, Minnesota. Transport Research Board, National 

Academy Press, Washington, D.C., pp. 178-186. 

Furniss, Michael J.; Staab, Brian P.; Hazelhurst, Sherry; 

Clifton, Cathrine F.; Roby, Kenneth B.; Ilhadrt, Bonnie 

L.; Larry, Elizabeth B.; Todd, Albert H.; Reid, Leslie 

M.; Hines, Sarah J.; Bennett, Karen A.; Luce, Charles 

H.; Edwards, Pamela J. 2010. Water, climate change, and 

forests: watershed stewardship for a changing climate. Gen. 


Tech. Rep.PNW-GTR-812. Portland, OR: U.S. Department 

of Agriculture, Forest Service, Pacific Northwest Research 

Station. 75 p. 

Gao, H., Q. Tang, X. Shi, C. Zhu, T. J. Bohn, F. Su, J. 

Sheffield, M. Pan, D. P. Lettenmaier, and E. F. Wood. 2010. 

Water Budget Record from Variable Infiltration Capacity 

(VIC) Model. In Algorithm Theoretical Basis Document for 

Terrestrial Water Cycle Data Records (in review). 

Gardali T, Seavy NE, DiGaudio RT, Comrack LA. 2012. 

A Climate Change Vulnerability Assessment of California's 

At-Risk Birds. PLoS ONE 7(3): e29507. doi:10.1371/journal. 

pone.0029507 

Glick, P., B.A. Stein, and N.A. Edelson, editors. 2011. 

Scanning the Conservation Horizon: A Guide to Climate 

Change Vulnerability Assessment. National Wildlife 

Federation, Washington, D.C. 

Hamlet, A. F., S. Lee, K. E. B. Mickelson, and M. M. Elsner. 

2009. Effects of projected climate change on energy supply 

and demand in the Pacific Northwest and Washington State, 

in The Washington Climate Change Impacts Assessment: 

Evaluating Washington’s Future in a Changing Climate, 

edited by J. S. Littell, M. M. Elsner, L. C. W. Binder and A.K. 

Snover, pp. 165-190 , University of Washington Climate 

Impacts Group, Seattle, WA. 

Hamlet, A. F., P. W. Mote, M. P. Clark, and D. P. Lettenmaier. 

2005. Effects of temperature and precipitation variability on 

snowpack trends in the western United States, J. Clim., 18, 

4545 4561. 

Holling, C. S. 1973. Resilience and stability of ecological 

systems. Annual Review of Ecology and Systematics 4: 1-23. 

Lee, D.C. and B.E. Rieman. 1997. Population viability 

assessment of salmonids by using probabilistic networks. 

North American Journal of Fisheries Management 

17:1144-1157. 

Luce, C. H., and Z. A. Holden. 2009. Declining annual 

streamflow distributions in the Pacific Northwest United 

States, 1948–2006, Geophys. Res. Lett., 36, L16401, 

doi:10.1029/2009GL039407. 

Luce, Charles; Morgan, Penny; Dwire, Kathleen; Isaak, 

Daniel; Holden, Zachary; Rieman, Bruce 2012. Climate 

change, forests, fire, water, and fish: Building resilient 

landscapes, streams, and managers. Gen. Tech. Rep. RMRS- 

GTR-290. Fort Collins, CO: U.S. Department of Agriculture, 

Forest Service, Rocky Mountain Research Station. 207 p.

Luo, L. F., and E. F. Wood. 2007. Monitoring and predicting 

the 2007 U.S. drought, Geophys. Res. Lett., 34, 6. 

Mote, P.W., A. F. Hamlet, M. P. Clark, and D. P. 

Lettenmaier. 2005. Declining mountain snowpack in western 

North America. Bull. Amer. Meteor. Soc., 86, 39–49. 

Ray, A.J., J.J. Barsugli, K.B. Averyt, K. Wolter, M. Hoerling, 

N. Doesken, B. Udall, R.S. Webb. 2008. Climate Change 

in Colorado: a Synthesis to Support Water Resources 

Management and Adaptation. Western Water Assessment. 

Boulder, CO. 

Rice, Janine; Tredennick, Andrew; Joyce, Linda A. 2012. 

Climate change on the Shoshone National Forest, Wyoming: a 

synthesis of past climate, climate projections, and ecosystem 

implications. Gen. Tech. Rep. RMRS-GTR-264. Fort Collins, 


CO: U.S. Department of Agriculture, Forest Service, Rocky 

Mountain Research Station. 60 p. 

Spears, M., L. Brekke, A. Harrison, and J Lyons. 2009. 

Literature Synthesis on Climate Change Implications for 

Reclamation’s Water Resources. Technical memorandum 

86-68210-091. U.S. Department of the Interior, Bureau of 

Reclamation, Research and Development Office. Denver, CO. 

USFS. 1995. Ecosystem Analysis at the Watershed Scale. 

Federal Guide for Watershed Analysis. USFS Northwest 

Region. Regional Ecosystem Office. Portland, Oregon. 26p. 

U.S. Department of Agriculture (USDA) Forest Service. 

2011a. Forest Service watershed condition classification 

technical guide. Washington, DC: U.S. Department of 

Agriculture, Forest Service, Watershed, Fish, Wildlife, Air, 

and Rare Plants Program. 

U.S. Department of Agriculture (USDA) Forest Service. 

2011b. Forest Service watershed condition classification

Pilot National Forest Reports 

Contents 

Gallatin National Forest ......................................................................... 30 

Helena National Forest ........................................................................... 46 

Grand Mesa, Uncompahgre and Gunnison National Forests ................ 64 

White River National Forest ................................................................ 112 

Coconino National Forest .................................................................... 130 

Sawtooth National Forest ..................................................................... 158 

Shasta Trinity National Forest ............................................................. 185 

Umatilla National Forest ...................................................................... 210 

Ouachita National Forest ...................................................................... 226 

Chequamegon-Nicolet National Forest ................................................ 236 

Chugach National Forest ...................................................................... 266 

28 Assessing the Vulnerability of Watersheds to Climate Change

Example of Recommended Citation Format for Forest Reports 

Caty Clifton; Day, Kate; Johnson, Allison. 2012. Assessment of Watershed Vulnerability to Climate Change, 

Umalilla National Forest. In: Michael J. Furniss, Roby, Ken B., Cenderelli, Dan; Chatel, John; Clifton, Caty F.; 

Clingenpeel, Alan; Hays, Polly E.; Higgins, Dale; Hodges, Ken; Howe, Carol; Jungst, Laura; Louie, Joan; Mai, S 

Christine; Martinez, Ralph; Overton, Kerry; Staab, Brian P.; Steinke, Rory; Weinhold, Mark. 2012. Assessing the 

Vulnerability of Watersheds to Climate Change: Results of National Forest Watershed Vulnerability 

Pilot Assessments. Climate Change Resource Center. U.S. Department of Agriculture, Forest Service 305p. 

www.fs.fed.us/ccrc/wva 


Assessment of Watershed Vulnerability 

to Climate Change 


April, 2012 

Prepared By: 

Joan Y. Louie 

Fisheries Biologist/GIS Analyst 

R1 Regional Office, Missoula, Montana 


Gallatin National Forest Watershed Vulnerability Assessment, Northern Region (R1) 

BACKGROUND 

The Gallatin National Forest (GNF) is located in southwestern Montana within the Northern Region (R1) 

of the U.S. Forest Service (USFS) and is part of the Greater Yellowstone Ecosystem, the largest intact 

ecosystem in the continental United States (Figure 1). The 1.8 million acre Forest contains more than 

1,900 miles of fish-bearing streams and 700 high mountain lakes, and supports important, high-profile 

recreational fisheries. 

Figure 1. The Gallatin National Forest is located in southwestern Montana, within the Greater 

Yellowstone Ecosystem. 



PARTNERS 

Data were provided by: 

• The Montana Natural Resource Information System (NRIS) 

• Montana Bureau of Mines and Geology (MBMG) 

• US Geology Survey (USGS) 

• University of Washington Climate Impacts Group (CIG) 

• Montana Fisheries Information System (MFISH) 

• US Forest Service (USFS) R1 Geospatial Group 

• Ecoshare 

Assistance with the analysis was provided by: 

• Kerry Overton et al., Rocky Mountain Research Station 

• Ralph Martinez, Plumas National Forest 

• Jim Morrison, R1 Regional Office 

ASSESSMENT OBJECTIVES 

The objective of this project was to develop a reliable method to prioritize all HUC-6 watersheds within 

the GNF in order to focus forest resource conservation and restoration efforts. A watershed 

characterization process was first developed to assess the relative sensitivity of the watersheds to 

disturbance, based on various environmental parameters. A vulnerability assessment further prioritized 

watersheds using the Watershed Condition Framework, resources of value, and exposure (climate 

projections). 

The proposed analysis has been developed in part to address the USFS initiative in considering climate 

change in land management decisions. Current studies show climate change is occurring, but climate 

model projections are uncertain and models at common management scales are nonexistent. Therefore, 

alternative methods of examining the potential impacts of climate change and other environmental 

stressors are needed. While this initial framework was originally designed from a watershed perspective, 

the results can also have implications for terrestrial management, such as fire, rangeland and wildlife 

management activities on the GNF. This process is intended to make it easy to update previous runs or 

examine other resources simply by rotating in the appropriate datasets. This project will also provide an 

example for other Forests in Region 1 to develop similar vulnerability assessments. 

SCALE OF ANALYSIS 

The scale of the analysis used in the GNF assessment was HUC-6 (12-digit) subwatersheds (Figure 2) and 

HUC-5 (10-digit) watersheds (Figure 3). 



Figure 2. Subwatersheds (HUC-6) on the Gallatin National Forest 

Figure 3. Watersheds (HUC-5) on the Gallatin National Forest 



CONNECTIONS TO OTHER PROJECTS AND ASSESSMENTS 

GNF Stream Temperature Modeling 

The objective of this project is to develop a broad-scale geographic information system (GIS) model to 

predict the effects of climate change on stream thermal regimes that in turn, provide the basis for 

estimating impacts on fisheries resources. A standardized approach was developed to collect and share 

temperature information among partner agencies and the public. The results will be used in conjunction 

with the GNF Watershed Vulnerability Assessment (WVA) to identify thermally-sensitive habitats and 

vulnerable native fish populations, and to prioritize future restoration activities to mitigate the effects of 

climate change on aquatic resources. 

GIS analysis identified locations for deployment of stream temperature loggers in HUC-6 watersheds 

intersecting the Gallatin and Custer National Forests. A matrix was developed comparing stream size (yaxis) 

and elevation (x-axis). Multiple temperature deployment locations were chosen from each cell of the 

matrix across broad spatial scales (see Figure 4 for the Lower East Boulder River HUC-6 watershed). 

Approximately 100 stream temperature loggers will be deployed, which include 40 long-term/multi-year 

deployments and 60 short-term/annual deployments. The data collected will be used to develop a model 

to predict changes in stream temperature with respect to elevation, contributing area (stream size), and air 

temperature. 

The methods employed were developed by the Rocky Mountain Research Station. For a complete 

description, refer to the following website. 

www.fs.fed.us/rm/boise/AWAE/projects/stream_temp/multregression/methods.shtml 

Blakely Creek 

East Boulder River 

Wright Gulch 

Canyon Creek 

Figure 4. Temperature deployment locations within the Lower East Boulder 

River HUC-6 watershed 

34 Assessing the Vulnerability of Watersheds to Climate Change 

Dry Fork Creek 

Burnt Gulch


Watershed Condition Framework 

The Watershed Condition Framework (WCF) was included as one step in the process. The WCF 

established a nationally consistent method for classifying watershed condition and documenting 

improvements in watershed condition at the forest, regional, and national scales (US Forest Service. 

2011). This process uses 12 indicators and 24 attributes to serve as surrogate variables representing 

fundamental ecological, hydrological, and geomorphic functions, and processes that affect watershed 

condition. The primary emphasis is on ecological processes and conditions that Forest Service 

management activities can influence. 

There are three watershed condition classes identified in this process: 

• Class 1 = Functioning Properly 

• Class 2 = Functioning at Risk 

• Class 3 = Functionally Impaired 

Watersheds considered to be Functioning Properly have ecosystem processes functioning within their 

range of natural variability. In general, the greater the departure from the natural pristine state, the more 

impaired the watershed condition is likely to be (USFS 2011). 

Climate Change Performance Scorecard 

The Climate Change Performance Scorecard is the Forest Service’s tracking tool to assess progress in 

integrating climate change considerations into programs, plans, and projects. It is composed of 10 

performance elements, with a national goal of 100% of Forests/Grasslands to achieve a “Yes” rating on 7 

of the 10 elements by FY 2015. One of these elements is a vulnerability assessment, which the WVA 

would fulfill. 

Forest Landscape and Rapid Assessments 

The WVA would not replace these assessments but can help validate priorities being identified in these 

assessments. 



WATERSHED VULNERABILITY ASSESSMENT PROCESS 

Figure 5. The Gallatin National Forest Watershed Vulnerability Assessment Model. The assessment consists of 

several different types of information (added and removed as necessary) to identify vulnerable watersheds. 

Geophysical/Sensitivity Characterization 

The first step of the WVA process, the geophysical/sensitivity characterization, was the most timeconsuming. 

As interdisciplinary team (fish biologist, hydrologist, and soil scientist), identified the 

dominant physical processes and features of the watershed that affect ecosystem function and condition. 

Identifying which watersheds are the most geophysically reactive can indicate how much a watershed 

responds to disturbances such as floods, drought, intense precipitation, and fires. The datasets determined 

to be most important for the watershed characterization were soils, geology, hydrology, terrain, and 

groundwater. 

The initial run of this analysis utilized pre-existing datasets (often outdated and of lower resolution and 

accuracy). These datasets include the GNF Soil Survey (slope classes, surficial geology, and shallow 

groundwater) and datasets derived from the National Hydrography Dataset, National Elevation Dataset, 

and R1 VMap (water yield, high flows, and low flows). After this initial run, the team met again to 

evaluate the results and determine which watershed characteristics were most important. 

The second run of the analysis included newer datasets developed for the analysis. The state surficial 

geology layer from MBMG was reclassified into broad rock class categories to identify sensitive 

geologies. A compound index of slope and aspect from 10m digital elevation models (USGS) was derived 

to identify sensitive terrain areas. The original hydrology metrics were omitted in the second run due to 

their strong correlation with the terrain analysis (see Hydrology section below). 

Each variable was quantified by subwatershed and given a rating of 1, 2, or 3, based on specific threshold 

values identified by literature and professional judgment. All scores were added together by 

subwatershed. Higher scores indicate higher sensitivity to disturbance. 

Geology Sensitivity 



The surficial geology layer covering the GNF was reclassified, based on a relative assessment of soil 

erosion/sediment delivery and rapid runoff potential from different bedrock types. Three classes were 

created (low, moderate, and high) to identify geology sensitivity by each HUC-6 subwatershed. 

Terrain Sensitivity 

A mathematical equation was used to explain the empirical relationship between slope, aspect, and 

elevation. The results of this analysis have been extrapolated beyond Forest boundaries to allow 

characterization of entire subwatersheds, however characterizations are truly only valid within Forest 

boundaries. The equation is developed for montane areas and will need to be recalibrated for use on flatter 

areas (outside of GNF boundaries). Three classes were created (low, moderate, and high) to identify 

terrain sensitivity for each HUC-6 subwatershed. A future iteration of this analysis will expand this 

terrain analysis beyond Forest boundaries to increase the accuracy. 

Geophysical Characterization 

The geology sensitivity and terrain sensitivity datasets were combined and reclassified with more weight 

given to the terrain dataset (Figure 6). 

Hydrology 

Groundwater is expected to play an important role in buffering the impacts of changing flows and stream 

temperatures, however currently there is no accurate and comprehensive dataset for groundwater. This 

information will be included in the model as better and more reliable methods of identifying groundwater 

data are determined. 

The first run of the WVA analysis developed hydrology metrics for water yield, high discharge, and low 

flows. Each metric was categorized into high, moderate, and low categories. The water yield sensitivity 

map compares reasonably well with the newly developed terrain sensitivity dataset. 

The main hydrology variable, water yield, appears to be accurately characterized, and is heavily 

influenced by the elevation variable. The aspect and slope steepness terrain variable further refines the 

elevation variable, accounting for less water yield on 150- to 210-degree aspect slopes and faster runoff 

on steep (35% + ) slopes. The hydrologic factors determined to be the most influential in watershed 

sensitivity to climate change are best represented by the terrain sensitivity analysis and, therefore, no 

hydrology metrics were included in the second run of the WVA. 



Figure 6. Geophysical characterization of Gallatin National Forest subwatersheds. 



Watershed Condition Framework 

The second step in the WVA assessment is the WCF dataset. This dataset identifies the level of human 

disturbance on the landscape. All of the GNF subwatersheds analyzed through this process were 

determined to be either Functioning Properly or Functioning at Risk (Figure 7). Because of this 

determination, some of the potentially more important watersheds may have been de-emphasized and 

future runs will need to confirm and/or modify this as needed. 

Figure 7. Watershed Condition Framework for the Gallatin National Forest 



Water Resources/Values 

The following water resources and values were chosen for the vulnerability analysis. 

1. Infrastructure 

• Roads 

• Trails 

• Developed recreation sites 

2. Water use and water developments 

• Point of use locations 

• Diversions 

3. Cutthroat Trout 

• “Sensitive species” designation by Forest Service 

• “Species of special concern” designation by state of Montana 

• Management indicator species for the GNF 

The purpose of this dataset is to quantify selected water resource values in each HUC-6 subwatershed. 

Areas with the greatest density of values may indicate important sites where there may have been 

significant economic investment and/or would require the greatest investment to maintain/conserve the 

resource. Datasets for each value were used to place subwatersheds into three classes (low, medium, and 

high) based on natural breaks in the data. All of the datasets were then combined to create one Values 

dataset, identifying subwatersheds with the lowest to highest amount of values (Figure 8). 

Figure 8. Levels of watershed resource/values by subwatershed. 



Exposure 

To evaluate exposure, we used the regional downscaled climate and hydrological projections developed 

by Littell et al. (2011), which build on research and data from the Climate Impacts Group (CIG) at the 

University of Washington. We chose their Ensemble model to examine potential climate change impacts 

more closely. This model is composed of the 10 best-fitting global circulation models (GCMs) for the 

Upper Missouri River Basin region. The modeled time periods available are 1916-2006 (historic), 2030- 

2049 (mid-21 st century), and 2070-2099 (late 21 st century). 

The climate projections are downscaled to 6 km 2 resolution and are most appropriately summarized at the 

HUC-5 scale. The HUC-6 subwatersheds were overlaid to examine how they may be influenced by these 

climate projections. The metrics retrieved for the most current run of the WVA include variable 

infiltration capacity (VIC) derived (Liang et al. 1994; Liang et al. 1996) hydrological projections: 

combined annual flow, seasonality of flow, and snowpack vulnerability (hydrologic regime). For the 

Upper Missouri River Basin, some of the overall trends predicted for the mid- to late 21 st century include 

increases in average annual air temperature, increases in seasonal air temperatures, increases in winter 

precipitation, and decreases in summer precipitation. 

Currently, we have used only the air temperature projections to examine predicted trends. In the future, 

these predicted air temperatures, combined with our stream temperature model (in development) and local 

air temperature data, may be used to model and predict stream temperatures across the forest. 

Potential Impacts to Water Resources 

1. Increased instances of low flows and lower flows 

• Water uses/diversions would amplify the anticipated low flows 

• Culverts currently passable by fish may become barriers during low flows 

2. Changes in flow regime 

• Increased winter flooding could increase summer low flows; increased/prolonged drought 

in the summer will further amplify the effects of changes in flow 

• Increased winter scouring of fall spawners (brook trout) 

− May favor native cutthroat trout 

3. Increased stream temperatures 

• Previously unsuitable stream habitats (too cold) may become suitable for fish 

• At lower elevations, native cold-water fish will be negatively affected 

− More tolerant invasive fish species may outcompete natives 

− 

4. Increased precipitation events 

• Roads would have increased sedimentation into streams 

• Culverts may need to be enlarged and/or maintained more frequently to accommodate 

higher flow 

• Some roads may need more frequent maintenance 

5. Increased drought events 

• Water use/diversions would exacerbate drought events 

• Possible increases in wildfires 



RESULTS 

Datasets for the geophysical characterization, Watershed Condition Framework, and water 

resources/values were overlaid for a composite result (Figure 9). The red and yellow subwatersheds 

indicate where our areas of interest have the most overlap. 

Figure 9. Composite result of the geophysical characterization, WCF, and water resources/values datasets 



SUMMARY 

The physical characterization of watersheds was the most time-consuming step of the WVA. We felt it 

was most important to first develop a robust analysis to characterize the geophysical reactivity of each 

subwatershed. We hoped to utilize the most up-to-date and readily available datasets. After our initial run 

using existing datasets of lower quality, we developed a repeatable method for the physical 

characterization of watersheds. The results appear to be reasonably accurate, although additional 

validations are needed. The terrain and geology sensitivity datasets may be used to derive other datasets, 

such as soils, for use in future iterations of this assessment and other Forest analyses. 

We expect this to be an iterative process that is never truly “complete.” The WVA was designed to be 

easily updated with the latest datasets as they are developed. This design allows different resource areas 

to be assessed together or separately, by incorporating the relevant datasets. Even as climate change 

projections are refined in the future, the physical characterization of our subwatersheds should remain the 

same, enabling quick evaluation of the subwatersheds through the latest climate scenarios without 

additional analysis. 

APPLICATIONS 

Management 

These results may aid GNF managers in prioritizing subwatersheds for resource conservation and 

restoration efforts. The results can also be used to validate priorities identified by the rapid assessments 

and landscape assessments on the Forest. 

Monitoring 

The identification of the potentially most sensitive and most vulnerable subwatersheds can be used to 

prompt monitoring in those areas at risk. 

Collaboration, Education and Outreach 

This analysis, and others like it, will hopefully provide more reason and opportunity for the USFS to 

educate the general public on climate change and our adaptive management strategies to address it. In 

addition, these analyses will provide opportunities to collaborate with other state, federal, and tribal 

agencies and non-governmental organizations (NGOs) to address climate change. 

CRITIQUE 

What important questions were not considered? 

Currently, the WVA should not be considered valid beyond the Forest boundaries. The terrain sensitivity 

analysis will need to be further refined to characterize the subwatersheds beyond the Forest boundaries. 

What were the most useful data sources? 

1. National datasets which do not end at the Forest boundaries (NHD, NED). 

− NED was very useful for deriving other datasets as well. 

2. Ecoshare (website) is a well-organized source for climate projections data for Region 1. 

3. Montana NRIS provided statewide datasets. 



4. Montana Bureau of Mines and Geology provided statewide surficial geology coverage. 

What were the most important data deficiencies? 

1. Data beyond forest boundaries. 

− The mathematical equation used to develop the terrain sensitivity analysis will theoretically 

work on any landscape, but currently has only been calibrated to a montane landscape. More 

effort will be needed to modify the equation and more accurately characterize entire 

subwatersheds that go beyond Forest boundaries. 

− The R1 VMap dataset may have great potential in future runs of the WVA; unfortunately, this 

is limited to the Forest boundaries and will likely stay that way. 

2. Groundwater data would be extremely helpful for this analysis, particularly to identify areas with 

buffering capacities to increased stream temperatures. Unfortunately, this data is currently lacking 

and it will be very time-consuming to develop an accurate dataset. 

3. Stream temperature data is also lacking on the GNF. We have only just begun a comprehensive 

effort in collecting this data, which, along with local air temperature data, will be helpful in the 

modeling and prediction of future stream temperatures. 

4. Field validations will be essential when there is available time and money. The physical 

characterization node of the WVA currently has only been “validated” by professional 

knowledge. 

What tools were most useful? 

1. ArcGIS – Without this program, spatial analyses would have been severely limited, particularly 

because open-source GIS programs are significantly less well-developed in user-friendliness, 

tools, and options. 

2. Google Earth is a useful tool to disseminate some of this spatial information for users who are not 

GIS-savvy. 

3. Video/phone conference calls, website and webinar technology greatly facilitated the group’s 

information-sharing and coordination, especially with limited funds for agency travel. 

What tools were most problematic? 

1. Citrix and T:\ drive on the Forest Service network. When fully functioning, these are excellent 

tools and make GIS more accessible for any Forest Service employee. Unfortunately, they have 

not yet reached their full potential and instead have created numerous issues for GIS users. 

2. ArcGIS often contains bugs and is not always the most intuitive for non-GIS people. New 

versions also come out relatively often and are mostly incompatible with the previous versions. 

This a non-issue for Forest Service employees utilizing Citrix, but can cause more issues when 

working with external agencies that cannot keep up with the latest ArcGIS versions. 

PROJECT TEAM 

Joan Louie, GIS analyst/fisheries biologist (R1 Regional Office) 

Scott Barndt, Forest fisheries biologist (GNF) 

Mark Story, Forest hydrologist (GNF) 

Tom Keck, Soil scientist (GNF) 



PROJECT CONTACT 

Joan Louie, GIS Analyst, R1 Regional Office 


Office: (406) 329-3209 

Email: joanlouie@fs.fed.us 

REFERENCES 

Littell, J.S., M.M. Elsner, G. S. Mauger, E. Lutz, A.F. Hamlet, and E. Salathe. 2011. Regional 

Climate and Hydrologic Change in the Northern US Rockies and Pacific Northwest: Internally Consistent 

Projections of Future Climate for Resource Management. Available online at: 

http://cses.washington.edu/picea/USFS/pub/Littell_etal_2010/ 

Liang, X., D. P. Lettenmaier, E. F. Wood, and S. J. Burges. 1994. A simple hydrologically based 

model of land-surface water and energy fluxes for general-circulation models, J. Geophys. Res.- 

Atmospheres, 99, 14415-14428. 

Liang, X., E. F. Wood, and D. P. Lettenmaier. 1996. Surface soil moisture parameterization of the 

VIC2L model: Evaluation and modification, Global Planet. Change, 13, 195–206. 

Rieman, B.E., Isaak, D.J. 2010. Climate Change, Aquatic Ecosystems, and Fishes in the Rocky 

Mountain West: Implications and Alternatives for Management. Gen. Tech. Rep. RMRS-GTR-250. Fort 

Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station. 46 p. 

US Forest Service. 2011. Watershed Condition Framework – A Framework for Assessing and Tracking 

Changes to Watershed Condition. FS-977. May 2011. http://www.fs.fed.us/publications/watershed/ 

10/3/11. 

Wenger, S.J., Isaak, D.J., Dunham, J.B., Fausch, K.D., Luce, C.H., Neville, H.M., Rieman, B.E., 

Young, M.K., Nagel, D.E., Horan, D.L., Chandler, G.L. 2011. Role of climate and invasive species in 

structuring trout distributions in the interior Columbia River Basin, USA. Can. J. Fish. Aquatic Sci. 68: 

988-1008. 

Wenger, S.J., Isaak, D.J., Luce, C.H., Neville, H.M., Fausch, K.D., Dunham, J.B., Dauwalter, D.C., 

Young, M.K., Elsner, M.M., Rieman, B.E., Hamlet, A.F., Williams, J.E. 2011. Flow regime, 

temperature, and biotic interactions drive differential declines of trout species under climate change. 

Proceedings of the National Academy of Sciences. 108: 14175-14180. 




Helena National Forest 

April, 2012 

Prepared By: 

Laura Jungst 

Hydrologist 

Helena National Forest, Helena, Montana 


Helena National Forest Watershed Vulnerability Assessment, Northern Region (R1) 

BACKGROUND AND FOREST CONTEXT 

The Helena National Forest is located in west-central Montana within the Northern Region (R1) of the 

USFS. The Forest consists of nearly 1 million acres of distinctive landscapes and lies on either side of the 

Continental Divide, resulting in a very diverse climate and landscape (Figure 1). The Forest’s watersheds 

make up the headwaters for both the Missouri and Columbia River basins. The western portion of the 

Forest straddles the Continental Divide starting at the southern tip of the Bob Marshall Wilderness and 

ending just east of Deer Lodge. The eastern side includes the lower, drier Big Belt Mountains. The Forest 

is composed of a mixture of grass and sagebrush covered lowlands with pockets of lodgepole pine and 

mountainous areas composed of Douglas fir, spruce and larch. Elevations do not exceed 10,000 feet 

(3,000 m). 

Figure 1. Helena National Forest (green) and nearby communities and rivers 

The Helena National Forest has a continental climate modified by the invasion of Pacific Ocean air 

masses. The Forest lies in the strong belt of westerly winds that move out of the Pacific Ocean and 

deposit much of their precipitation on the mountain ranges in western Montana. Summers are warm in 

most valleys and cooler in the mountains. Winter months are relatively cold. Most precipitation falls as 

snow, and a deep snowpack accumulates in the mountains. East of the Continental Divide, occasional 

down slope warming winds, Chinooks, can occur in the winter months, resulting in a rapid rise in air 

temperature. The average annual precipitation ranges from 11.21 inches at Townsend in an intermountain 

valley to 50.30 inches at Copper Creek on an alpine mountain ridge. Valleys generally receive two-thirds 

to three-fourths of their annual precipitation during the growing season with seasonal peaks in May and 

June and again in September. The mountainous areas receive a larger percentage of their precipitation as 

snow during the winter. Average annual snowfall varies from 30 inches at Holter Dam to 108 inches at 

Lincoln Ranger Station (Sirucek, 2001). 

ANALYSIS OVERVIEW 

The WVA was completed for all subwatersheds under the management of the Helena National Forest. 

Three steps were completed to determine the vulnerability of each subwatershed (Hydrologic Unit Code 

level 6 (HUC-6)) to predicted changes in climate. First, the sensitivity of each subwatershed was 

determined, based on existing data representing the current condition of the subwatershed for each 

individual resource value of concern. Next, an exposure analysis was conducted based on the selected 



climate variables assigned to each resource value. Lastly the sensitivity analysis outcome was overlaid 

with the exposure analysis outcome to show final watershed vulnerability for each HUC-6. 

Several different analysis units were used as part of this assessment. Sensitivity analysis was summarized 

at the subwatershed level as delineated by the sixth level (12-digit) hydrologic unit (HUC-6) hierarchy in 

the US Geological Survey (USGS) National Hydrography Dataset (NHD). Because many of the forest 

management decisions and projects are conducted at the subwatershed scale or smaller, we chose to use 

this scale to make this analysis most useful on the ground. This analysis includes 151 subwatersheds 

within the assessment area. 

The exposure analysis was conducted at the watershed scale (HUC-5) (Figure 2). This scale was used 

because the climate data was downscaled to around a 6 km hydrologic output; this data fit our analysis 

best at the HUC-5 watershed level. 

To resolve these differences in scale, we used the sensitivity analysis at the subwatershed scale and 

overlaid climate predications at the watershed scale to show how underlying subwatersheds may be 

influenced by the climate predictions, while keeping the focus at a reasonable management scale. 

Figure 2. HUC-6 subwatersheds and HUC-5 watersheds within Helena National Forest 

WATER RESOURCE VALUES 

The following water resource values were chosen for the vulnerability analysis. Although there are many 

water resource values on the Helena National Forest, we analyzed the three values that we believe are of 

greatest concern to the Forest. 



Bull trout 

• Listed as a Threatened Species throughout their range under the Endangered Species Act since 

1999. 

• Have important habitat on the Helena National Forest west of the continental divide in the 

headwaters of the Columbia River. 

• Require colder water temperatures than most salmonids. 

• Require the cleanest stream substrates for spawning and rearing. 

• Need complex habitats, including streams with riffles and deep pools, undercut banks, and lots of 

large logs. 

• Rely on river, lake, and ocean habitats that connect to headwater streams for annual spawning and 

feeding migrations. 

Cutthroat trout 

• One of two subspecies of native cutthroat found in Montana. 

• Montana’s state fish. 

• Historic range was west of the Continental Divide as well as the upper Missouri River drainage. 

• Range has been seriously reduced due to hybridization with rainbow and/or Yellowstone 

cutthroat and habitat loss and degradation. 

• Designated a Montana Fish of Special Concern in Montana. 

• Common in both headwaters lake and stream environments. 

Infrastructure 

• Roads, campgrounds near streams and rivers, water diversions, bridges, etc. 

• Can become a safety concern for all forest users recreating in areas where streams are subject to 

higher flows, flash floods, etc. 

• Important financial investment for the Forest Service. 

EXPOSURE 

Information on predicted climate changes anticipated on the Helena National Forest came from a variety 

of sources. Published reports from the Rocky Mountain Research station were used to describe the 

general projections for the region including the projected change in the climate variable, the anticipated 

watershed response, and the potential consequences to watershed services (Table 1) (Rieman and Isaak, 

2010). Generally, predictions agree on a warmer and sometimes drier climate (Rieman and Isaak, 2010). 

This will include an increase in summer maximum temperatures of approximately 3 °C by the mid-21 st 

century, and an increase in spring and summer precipitation accompanied by a decrease in fall and winter 

precipitation. 



Projected Climatic 

Changes 

Anticipated Watershed 

Response 

Warmer air temperatures • Warmer water 

temperature in streams 

Changes in precipitation 

amounts and timing 

Less snowfall, earlier 

snowmelt, increased 

snowpack density 

Intensified storms, 

greater extremes of 

precipitation and wind 

• Altered timing and 

volume of runoff 

• Altered erosion rates 

• Higher winter flows 

• Lower summer flows 

• Earlier and smaller peak 

flows in spring 

• Greater likelihood of 

flooding 

• Increased erosion rates 

and sediment yields 


Potential Consequences to Watershed 

Services 

• Decrease in coldwater aquatic habitats 

• Increases or decreases in availability of 

water supplies 

• Complex changes in water quality related 

to flow and sediment changes 

• Changes in the amounts, quality and 

distribution of aquatic and riparian 

habitats and biota 

• Changes in aquatic and riparian habitats 

• Increased damage to roads, campgrounds, 

and other facilities 

Table 1. Projected hydrologic changes relative to the HNF identified values. Adapted from Water, Climate Change, 

and Forests GTR (Rieman and Isaak, 2010) 

The models used to predict climate changes were developed by the Climate Impacts Group (CIG) at the 

University of Washington. The Climate Impacts Group selected the A1B climate scenario to provide 

projections most relevant for vulnerability assessment and scenario planning exercises. They then 

modeled change (from time period 1916-2006 representing historic) and for two future time periods 

representing the mid-21 st century (2030-2049) and late 21 st century (2070-2099), using the emissions 

scenario A1B with the composite climate model. The composite model is an ensemble of climate models 

that falls between those models that predict cooler and warmer climate scenarios. It includes 10 Global 

Circulation Models that perform similarly well in the PNW / Columbia Basin, the Northern Rockies / 

Upper Missouri Basin, and the Central Rockies / Upper Colorado Basin and this is what the Helena 

National Forest chose to use to represent climate change in this analysis. Data was summarized at the 

HUC-5 scale for the entire Forest (downloaded from ftp://ftp2.fs.fed.us/incoming/gis/PNF/WVA/ on 

12/10/2010). 

Predicted changes in selected hydrologic attributes were derived from the Variable Infiltration Capacity 

(VIC) model. Parameters from VIC modeling were used to assess potential impacts to the selected forest 

water resource values. We compared the HUC-5 scale CIG’s VIC outputs for the historic trend and 

composite models for the following parameters (by resource value): 

1. Bull trout – Average summer maximum air temperature 

2. Cutthroat trout – Average summer maximum air temperature 

3. Infrastructure – Snowpack vulnerability (defined as the ratio of April1 snow water equivalent and 

October-March precipitation)


Predicted Average Summer Maximum Temperature 

Projection Period Historic Composite 

Mean Range Mean Range 

1916-‐2006 23.7 21.1-‐26.0 

2030-‐2059 26.0 (+10%) 23.4-‐28.2 

2070-‐2099 28.2 (+19%) 25.6-‐30.5 

Predicted Average April 1 SWE 



1916-‐2006 41.3 0.2-‐342.7 

2030-‐2059 29.6 (-‐28%) 0-‐289.2 

2070-‐2099 18.6 (-‐55%) 0-‐216.9 

Predicted Average June Runoff 



1916-‐2006 10.1 2.9-‐87.6 

2030-‐2059 7.4 (-‐27%) 2.5-‐59.4 

2070-‐2099 5.6 (-‐45%) 2.3-‐33.0 

Summer Baserflow (September Runoff) 



1916-‐2006 2.3 1.1-‐5.2 

2030-‐2059 2.5 (9%) 1.1-‐7.0 

2070-‐2099 1.2 (-‐48%) 0.2-‐3.2 

Table 2. Historic (1916-2006) and future (2030-2059 and 2070-2099) hydrologic output climate predictions 

averaged over all watersheds on the Helena National Forest. Based on Global models downscaled to 1/16th degree 

(~6 km) grid. 

VULNERABILITY ANALYSIS BY RESOURCE VALUE 

For this watershed vulnerability assessment, pilot forests were tasked with identifying the relative 

vulnerability of watersheds to potential risks posed by climate change by focusing on the potential effects 

of those changes to water resource values. Based on our current evaluation of water resource values on 

the Helena National Forest, values evaluated include fisheries habitat for bull trout and cutthroat trout, 

and infrastructure. Vulnerability analysis was conducted specific to each individual water resource value. 

Water Resource Value: Bull Trout Habitat 

Sensitivity 

Bull trout habitat condition was characterized using the regional bull trout watershed baseline analysis 

completed in 2007. This analysis was a consultation requirement for species listed under the Endangered 

Species Act since the late 1990’s. Baseline information was summarized according to important 

environmental parameters for each subwatershed within the Helena National Forest. This summary was 



divided into six overall pathways (Table 3). Each of the pathways is categorized in terms of functionality; 

either Functioning Appropriately (FA), Functioning at Risk (FAR), or Functioning at Unacceptable Risk 

(FUR). The final rating is based on a suite of metrics which are either (1) quantitative metrics of collected 

field data or GIS driven attributes (e.g. road density) or (2) qualitative descriptions based on field reviews, 

professional judgment, etc. 

The composite watershed sensitivity based on the baseline analysis is depicted in Figure 3. Based on these 

parameters, the Helena National Forest has three subwatersheds rated as FA, ten rated as FAR and five 

rated as FUR. These rating is applied to only those subwatersheds where there are known populations of 

bull trout. Evaluation of those watersheds that could have potential for bull trout habitat but do not 

currently have viable populations were not included in this analysis. 

Exposure 

Summer maximum air temperature predictions were used as a surrogate for stream temperature because 

stream temperature data was not widely available. At the time of our analysis this was our best available 

dataset, in the future, it might be better to use mean summer temperature as better correlations have been 

found between air-water temperatures using the mean vs. max, even though these were very strongly 

correlated (Wenger et al. 2011a). Summer maximum air temperatures were predicted to increase by 

approximately 2 °C uniformly across the forest for the 2030-2059 predictive period and approximately 5 

°C uniformly for the 2070-2099 predictive period. Consequently, it is predicted that not any one 

watershed will be more impacted by this change in summer maximum air temperature than another. 

However, we can develop conservation strategies based on current conditions in order to buffer more 

highly valued watersheds. 

Summer baseflow was considered as an exposure element, but not carried forward because Wenger’s 

(2011a) work showed temperature to be the key climate change variable related to bull trout habitat. Bull 

trout are likely sensitive to increase in winter high flows (Wenger 2011b), but this data is available at the 

reach level and time at this point does not allow for this kind of analysis. Winter 95 represents the number 

of days during winter that are among the highest 5% (respectively) of flows for the year. Winter 95 was 

used as the variable for winter high flows which would affect bull trout and brook trout, but not the spring 

spawning Westslope cutthroat trout. 

Watershed Vulnerability 

By overlaying the climate exposure data to the bull trout fisheries baseline data we see which habitat 

currently supporting bull trout populations is most likely to be adversely impacted by changes such as 

increased temperatures. Research has found bull trout currently inhabit the coldest available headwater 

streams which leaves little potential to shift to higher elevation habitats to avoid temperature increases 

(Wenger 2011a). Because the predicted temperature changes on the Helena National Forest are very 

uniform across all bull trout habitat, we assumed that it all has similar potential to be impacted by changes 

in climate. However, forest managers have the capability to maintain or increase the resiliency of 

watersheds that support the most valued bull trout fisheries. These areas can be selected as high priority 

for management activities. Because exposure to increased air temperatures is essentially uniform across 

the Forest, composite watershed vulnerability for bull trout habitat is equal to the watershed sensitivity 

analysis (Figure 3) or the current condition of the fisheries habitat. Incorporation of other climate change 

indicators may or may not change the overall potential vulnerability of these watersheds. 



Category Metric 

Subpopulation Characteristics within 

Subpopulation Watersheds 

Subpopulation size 

Growth and survival 

Life history diversity and isolation 

Persistence and genetic integrity 

Temperature 

Habitat - Water Quality 

Sediment 

Chemical contamination/nutrients 

Habitat - Access Physical barriers 

Substrate embeddedness in rearing areas 

Large woody debris 

Pool frequency and quality 

Habitat - Elements 

Large pools 

Off-channel habitat 

Refugia 

Average wetted width/maximum depth 

Ratio in scour pools in a reach 

Channel Condition and Dynamics 

Streambank condition 

Floodplain connectivity 

Change in peak/base flows 

Flow/Hydrology 

Increase in drainage network 

Road density and location 

Disturbance history 

Watershed Conditions 

Riparian conservation areas 

Disturbance regime 

Table 3. Matrix of Pathways and Indicators 



Figure 3. Composite watershed sensitivity rating for bull trout 



Water Resource Value: Westslope Cutthroat Trout Habitat 

Sensitivity 

Cutthroat trout habitat was assessed using the “Matrix of Pathways and Indicators” for bull trout fisheries 

described above (Table 3) in combination with the cutthroat distribution information and professional 

knowledge of the specific subwatersheds. The ratings of FA, FAR and FUR were given to those 

watersheds with known populations of cutthroat trout. Of the 155 watersheds with some portion of their 

area under forest management, 76 have known populations of cutthroat trout. Five of these have a rating 

of FA, 19 are FAR, and 52 are FUR (Figure 4). 

Exposure 

Westslope cutthroat trout are closely associated with headwater habitats which are often more stable than 

downstream reaches. Therefore, they may be less influenced by changes in large scale environmental 

conditions (Copeland, 2011, Wenger 2011a). Cutthroat trouts have a strong negative response to brook 

trout presence at the subwatershed scale (Wenger, 2011b). Brook trout are highly sensitive to increasing 

temperature, so the cutthroat trout could have an indirect positive response to climate change (Wenger, 

2011b). Most of the subwatersheds on the Helena National Forest have a population of invasive brook 

trout, brown trout or both. Only 17 of the subwatersheds do not have known populations of these invasive 

species (Figure 5). We analyzed the effects of average summer maximum temperature increase as a net 

positive interaction with cutthroat trout due to the parameter’s negative interaction with the invasive 

populations of brook trout (Wenger, 2011b). 

Figure 4. Composite watershed sensitivity rating for Westslope cutthroat trout 



Composite Watershed Vulnerability 

Cutthroat trout will experience the same changes in temperature that bull trout experience because they 

too inhabit headwater streams. Summer maximum air temperatures were predicted to increase by 

approximately 2 °C uniformly across the forest for the 2030-2059 predictive period and approximately 5 

°C uniformly for the 2070-2099 predictive period. These temperature changes are assumed to have little 

impact on cutthroat populations; however there could be a net positive effect due to the predicted decrease 

in invasive populations (Figure 5). 

Figure 5. Composite watershed sensitivity rating for Westslope cutthroat trout with invasive fish species (bull trout 

and brown trout) habitat overlay 

Water Resource Value: Infrastructure 

Sensitivity 

Based on the indicators used to determine sensitivity (Table 5), a rank was developed to show those 

watersheds that are least resilient (most sensitive). The overall sensitivity score was determined by 

calculating the average of the ranked values given to each of the sensitivity factor. Density of high value 

near-stream developments (table 5) were used to characterize infrastructure value (Figure 6). 



Resource Value at Risk Description 

Water diversions 

Number of diversions per subwatershed. Based on 

Montana State water rights and diversion data. 

Municipal Watershed Current municipal water use. 

Recreation Developments 

Developed recreation sites within 200' of stream (i.e., 

campgrounds, picnic grounds, trailheads, etc.) 

Riparian roads Roads miles within 150' of a stream. 

Sensitivity Factor Description 

Number of Road/Stream Crossings 

Stream crossings were determined by intersecting 

perennial and intermittent streams with existing roads. 

Percent severe and/or moderate erosion potential 

Soils 

determined using the erosion potential designated by the 

Helena National Forest Soil Survey. 

Roads Road miles by subwatershed. 

Riparian roads Roads miles within 150' of a stream. 

Table 5. Resource values considered and indicators used to determine infrastructure sensitivity 

Figure 6. Sensitivity ratings for the infrastructure value on the Helena National Forest. Map highlights 

Tenmile watershed, a watershed with high sensitivity due to its function as a municipal watershed. 



Exposure 

Many parameters influence the timing and magnitude of runoff for a given watershed. Average winter 

precipitation and average maximum winter temperature were initially analyzed to determine watershed 

exposure to changes in climate variables. 

Precipitation average over the entire forest is projected to increase slightly. All elevations on the forest are 

above approximately 3,500 feet. The average elevation is approximately 6,200 feet and the maximum 

elevation approximately 9,500 feet. Precipitation is predicted to increase in the winter, spring and fall and 

decrease in the summer season. 

Projected maximum winter temperatures (Dec-Jan-Feb) for the Helena National Forest for the 2040s time 

period are expected to increase. The average temperature across all HUCs went from 0 °C historically to 

1.3 °C projected for the 2040s time period. Temperatures are expected to remain relatively cold with the 

average maximum winter temperature not exceeding 3 °C for any individual watershed. Temperature is 

predicted to continue to increase into the 2080s time period where the average maximum winter 

temperature for all watersheds is predicted to be near 3 °C. Hamlet and Lettenamaier, 2007, found 

through a series of models of the northwestern United States, that cold river basins, where snow processes 

dominate the annual hydrologic cycle (< 6 °C average in midwinter), typically show reductions in flood 

risk due to overall reductions in spring snowpack. The Helena National Forest is well below 6 °C average 

midwinter and may see reductions in spring runoff flows for this reason. 

Since changes in summer and winter temperature are not expected to have a direct effect on infrastructure 

and development, change in watershed snowpack (the ratio of April 1 st SWE to Oct-Mar precipitation) 

was the only climate factor used to assess exposure. This value has been calculated using downscaled 

climate and hydrologic projections for the entire Columbia, upper Missouri and upper Colorado basins. 

Figures 7 and 8 show predicted watershed snowpack vulnerability (Littell et al.) watershed for the 2030- 

2059 and 2070-2099 time periods, respectively. Both the North Fork of the Blackfoot River and The 

Landers Fork watersheds are projected to see the most change in snowpack. 



Figure 7. Changes in rainfall/snowmelt dominance for HUC-5 watersheds on the 

Helena National Forest predicted for the 2030-2059 time period 

Figure 8. Changes in rainfall/snowmelt dominance for HUC-5 watershed on the 

Helena National Forest predicted for the 2070-2099 time period 




Watershed vulnerability was determined by overlaying the sensitivity analysis with the exposure analysis. 

Watersheds with include high value infrastructure and high sensitivity that also had highest risk of 

snowpack loss where rated as most vulnerable. The most vulnerable watersheds (Figures 9 and 10) are 

found in the northernmost section of the forest where changes in winter snowpack possibly resulting in 

rain on snow events pose the highest risk to forest infrastructure. 

Figure 9. Watershed vulnerability with regards to forest infrastructure is based on 

watershed sensitivity and exposure results for the 2030-2059 time periods 



Figure 10. Watershed vulnerability with regards to forest infrastructure is based on watershed 

sensitivity and exposure results for the 2070-2099 time periods 

CONNECTIONS TO OTHER ASSESSMENTS AND POTENTIAL APPLICATIONS 

• The WVA will provide a basis for incorporating climate change considerations into project 

planning and implementation. Identified climate change considerations may also be designed into 

forest plan desired conditions, objectives, and standards and guidelines. 

• Information from the WVA, while not specifically part of the watershed condition framework, 

can be used to help identify priority watersheds for future restoration activities. 

• Completing the WVA will aide in the completion of the climate change scorecard. The WVA 

analysis helps fulfill element 6 (vulnerability assessment), element 7 (adaptation activities), and 

element 8 (monitoring). 

• The WVA utilized work done by the Fisheries Watershed Baseline for the bull trout and 

Cutthroat trout sensitivity analysis. 

CRITIQUE 


1. The watershed vulnerability assessment focused only on water resources and did not consider 

predicted changes to terrestrial resources. While this analysis was designed to focus on water 

resources, composite effects on terrestrial ecosystems can have significant influence on watershed 

hydrology. 

2. Did not account for all resilience factors and did not use all climate exposure factors, including 

flow metrics. 




1. The Forest Service GIS database as well as the state GIS database (NRIS) was useful in 

describing sensitivity on a watershed basis (http://nris.mt.gov/). 

2. VIC data available from the Climate Impacts group was useful in describing projected climate 

change under several models. 


1. The data analyzed was based on layers that were approximations of what is on the ground. For 

example, NHD Streams and the roads layers are approximations and resulting stream crossing 

point layer is not necessarily an accurate representation. Field inventories in general are not 

complete. This is a data gap that could be improved in the future. 

2. Climate data was complex and time consuming to use. 


1. Examples of how the analysis was approached on other units including what kind of data to 

include and how to organize and display the information. 

2. Communication and support from all members of the WVA group willing to share their ideas and 

experiences throughout the process. Information sharing included monthly conference calls and 

Google share site. 

3. ArcGIS was a necessary tool throughout the entire process including evaluation and display of all 

data. 

4. Microsoft Excel was used as an interface between tabular data and spatial data. Often tables 

would be exported from ArcGIS to excel, manipulated and then imported and new values could 

then be displayed spatially. 


1. Downscaled climate data 

2. Forest level GIS data 

3. Accurately displaying climate change projections and resolving differences in scale between the 

forest level data and downscaled climate data. 

PROJECT TEAM 

Core Team: Laura Jungst, Hydrologist; Dave Callery, Hydrologist; Len Walch, Fisheries Biologist 

Support: Melanie Scott, GIS analyst; Kerry Overton, RMRS 

Data: Climate Impacts Group (Variable Infiltration Capacity (VIC) modeled data for several climate 

change scenarios at the HUC-5 scale and raster data at the 6 km grid scale) 

• RMRS – Boise, Kerry Overton 

• Jim Morrisson 

• Montana Natural Resource Information System Digital Atlas of Montana 

(http://maps2.nris.mt.gov/mapper/) 

• Helena National Forest GIS analyst Melanie Scott 



• Western US Stream Flow Metric Dataset 

(http://www.fs.fed.us/rm/boise/AWAE/projects/modeled_stream_flow_metrics.shtml) 

• Helena National Forest Fisheries Watershed Baseline 


Laura Jungst, Hydrologist 

Helena National Forest 

ljungst@fs.fed.us 

(406) 495-3723 

REFERENCES 

Copeland, T., Meyer, K.A. 2011. Interspecies Synchrony in Salmonid Densities Associated with Large- 

Scale Bioclimatic Conditions in Central Idaho, Transactions of the American Fisheries Society, 140:4, 

928-942 

Hamlet, A.F., Lettenmaier, D.P. 2007. Effects of 20 th century warming and climate variability on flood 

risk in the western U.S. Water Resources Research, 43, W06427. 

Sirucek, D., 2001. Soil Survey of the Helena National Forest Area, Montana. USDA Forest Service and 

natural Resources Conservation Service. Northern Region. 

Littell, J.S., M.M. Elsner, G. S. Mauger, E. Lutz, A.F. Hamlet, and E. Salathé. 2011. Regional 

Climate and Hydrologic Change in the Northern US Rockies and Pacific Northwest: Internally Consistent 

Projections of Future Climate for Resource Management. Draft report: January 7, 2011. Online at: 

http://cses.washington.edu/picea/USFS/pub/Littell_etal_2010/ 

Wenger, S.J., Isaak, D.J., Luce, C.H., Neville, H.M., Fausch, K.D., Dunham, J.B., Dauwalter, D.C., 

Young, M.K., Elsner, M.M., Rieman, B.E., Hamlet, A.F., Williams, J.E. 2011(a). Flow regime, 

temperature, and biotic interactions drive differential declines of trout species under climate change. 

Proceedings of the National Academy of Sciences. 108: 14175-14180. 

Wenger, S.J., Isaak, D.J., Dunham, J.B., Fausch, K.D., Luce, C.H., Neville, H.M., Rieman, B.E., 

Young, M.K., Nagel, D.E., Horan, D.L., Chandler, G.L. 2011(b). Role of climate and invasive species 

in structuring trout distributions in the interior Columbia River Basin, USA. Can. J. Fish. Aquatic Sci. 68: 

988-1008. 

Rieman, B.E., Isaak, D.J. 2010. Climate Change, Aquatic Ecosystems, and Fishes in the Rocky 

Mountain West: Implications and Alternatives for Management. Gen. Tech. Rep. RMRS-GTR-250. Fort 

Collins, CO: U.S. Department of Agriculture, Forest Service, Rocky Mountain Research Station. 46 p. 




Grand Mesa, Uncompahgre and Gunnison 

National Forests 

March, 2012 

Prepared by: 

Carol S. Howe 

Resource Information Specialist, Climate Change Coordinator 

and John Almy, Clay Speas, Warren Young and Ben Stratton, 

Grand Mesa, Uncompahgre and Gunnison 

National Forests, Delta, Colorado 


Grand Mesa, Uncompahgre and Gunnison National Forest Watershed Vulnerability Assessment, Rocky 

Mountain Region (R2) 

LOCATION 

The Grand Mesa, Uncompahgre, and Gunnison National Forests (GMUG) are located in western 

Colorado (Figure 1), within the Rocky Mountain Region (R2) of the USFS. 

Figure 1. Grand Mesa, Uncompahgre, and Gunnison National Forest vicinity map 

The GMUG is also located within the headwaters of the Upper Colorado River Basin (Figure 2). 




PARTNERS 

Figure 2. Upper Colorado River Basin 

Data were obtained from the following groups: 

• Rocky Mountain Research Station (facilitated training on use of climate tools, developed climate 

record for GMUG (pending), PRISM data, from Linda Joyce, Chuck Rhoades, David Coulson) 

• Western Water Assessment (WWA) (climate data websites from Jeff Lucas) 

• The Nature Conservancy (climate change scenarios for the Gunnison Basin, prepared by Joe 

Barsugli (WWA) and Linda Mearns (National Center for Atmospheric Research)) 

• Climate Impacts Group (Variable Infiltration Capacity (VIC) modeled data for several climate 

change scenarios at the HUC-5 scale and raster data at the 6 km-grid scale) 




• Colorado Department of Water Resources (surface and groundwater sources, water rights 

information) 

• Colorado Department of Public Health and Environment (source water protection areas) 

SCALES OF ANALYSIS 

Area Assessed 

The WVA was completed for the entire Forest and surrounding areas, in general; and specifically for 

those portions of the GMUG within watersheds that were mostly on National Forest System lands. 

Analysis Units 

Several different analysis units were used as part of this WVA. Analyses were summarized primarily at 

the subwatershed level (-6), as delineated by the sixth level (12-digit) of the hydrologic unit hierarchy in 

the US Geological Survey (USGS) National Hydrography Dataset (NHD) and the US Department of 

Agriculture (USDA) Natural Resources Conservation Service (NRCS) Watershed Boundary Dataset 

(WBD). There are 205 subwatersheds within the assessment area for this WVA. 

Some subwatersheds were merged together for analysis purposes so that complete catchment basins were 

delineated (some HUC-6 subwatershed delineations from NHD/WBD separated upper portions of 

watersheds from lower portions). This resulted in 152 modified HUC-6 subwatersheds. These modified 

HUC-6 subwatershed analysis units were used to summarize information on aquatic resource values, and 

watershed risks described below as inherent sensitivities and anthropogenic stressors. 

Anticipated climate changes, or exposure (also described below) were evaluated using several different 

analysis units. Watersheds (HUC-5), delineated at the fifth level (10-digit) of the hydrologic unit 

hierarchy in NHD/WBD were used to summarize predicted climate changes output by the VIC model. 

There are 49 HUC-5 watersheds that overlap the assessment area for this WVA. 

In addition, exposure was also evaluated using geographic areas that have similar climatic regimes. These 

geographic areas also roughly correspond to areas used in Forest planning. Modified HUC-6 

subwatersheds were aggregated into six geographic areas within the assessment area. 

Figure 3 shows the original NHD/WBD HUC-6 delineations, the modified HUC-6s used for this analysis, 

and the HUC-5 watersheds as they overlap the GMUG. Figure 4 shows the geographic overlap of the 

modified HUC-6 subwatersheds and the GMUG. 




Figure 3. Modified HUC-6 Subwatersheds and HUC-5 Watersheds used in Watershed Vulnerability Assessment 

Uncompahgre 

Grand Mesa 

San Juans 

West Elks 

Upper Taylor 

Cochetopa 

Figure 4. Geographic Area and Modified HUC-6 Subwatersheds used in Watershed Vulnerability Assessment 




CONNECTIONS TO OTHER ASSESSMENTS, PLANS AND EFFORTS 

The WVA used data and results from a previous watershed assessment completed as part of the 2005- 

2007 Forest plan revision process, specifically: 1) a summary of past activities that have occurred in each 

subwatershed (used as the anthropogenic stressors in the WVA); 2) a summary of intrinsic characteristics 

of each subwatershed (i.e. geology, soil types, topography) that indicate how sensitive a given watershed 

is to erosion (used as the indicator for erosion and sediment production for the WVA); and 3) a summary 

of water uses by subwatershed (used as the water uses values for this WVA). Data and results from the 

Forest plan watershed assessment were limited to National Forest System lands. Off-Forest data were 

lacking or very limited and were not incorporated into the existing data for the WVA. The WVA will 

incorporate consideration of potential effects of predicted climate changes, which was not previously 

done. 

Results of the WVA will be used as part of a vulnerability assessment for the Upper Gunnison Basin, an 

ongoing collaborative effort with The Nature Conservancy (part of its Southwest Climate Change 

Initiative), the BLM, National Park Service, Gunnison County, Colorado Division of Wildlife, Colorado 

River Conservation Board and the USFS. The Upper Gunnison Basin vulnerability assessment will 

incorporate terrestrial resources that the WVA did not, as well as aquatic resources that occur off the 

National Forest. 

The WVA will also inform additional outcomes from the Upper Gunnison Basin collaborative effort, 

which include: 1) developing landscape-scale strategic guidance for climate adaptation and resiliencebuilding 

for a set of conservation targets (e.g., Gunnison sage-grouse); 2) developing tools and 

information to make current conservation projects climate smart; and 3) developing a climate adaptation 

demonstration project. 

The WVA and the subsequent vulnerability assessment for the Gunnison Basin will provide a basis for 

incorporating climate change considerations into project planning and implementation. When Forest plan 

revision efforts resume on the GMUG, identified climate change considerations can also be designed into 

Forest plan desired conditions, objectives, standards, and guidelines. 

Data gaps and uncertainties in predicting climate changes and potential effects are needs that can be filled 

through a variety of monitoring efforts. 

In 2011, the GMUG NF completed a Watershed Condition Classification. Information from the WVA, 

while not specifically part of the watershed condition classification protocol, can be used to help identify 

priority watersheds for future restoration activities. 

WATER RESOURCES 

This WVA is intended to identify the relative vulnerability of watersheds to potential risks posed by 

climate change, by focusing on the potential effects of those changes to water resource values. For the 

pilot project, water resource values needed to include floodplain and in-channel infrastructure, water uses, 

and aquatic species. Following this direction, the GMUG team initially identified a list of resources in 

these three categories. As we worked through the process, lack of available data and time constraints 

reduced the list of values that were ultimately evaluated. We also adjusted how several resource values 

were grouped so that the final three categories combined values that would respond in similar ways to 

predicted climate changes. Modifications made during the process are discussed for each category, below. 




Infrastructure Values 

Infrastructure includes roads, trails, culverts, bridges, recreation developments, and other structures that 

have been constructed for some purpose. Infrastructure associated with stream channels, floodplains, and 

riparian areas was believed to be most vulnerable to changes in timing or magnitude of stream flow. The 

NHD flowline data were used to identify stream courses. Floodplains and riparian areas were identified 

using the Forest riparian habitat layer (created from aerial photo interpretation to identify wetlands, fens, 

waterbodies, and 100-foot buffer of perennial streams). 

The infrastructure values evaluated in this WVA are listed below, along with the metric used to rank these 

values by watershed. Note: Data for riparian areas, roads and trails, recreation developments, and 

recreation residences were limited to National Forest System lands and what is available in USFS 

databases. Stream network information extended off-Forest. The discrepancies in data extent means the 

confidence in results varies for subwatersheds that are completely or mostly within the GMUG as 

compared to those subwatersheds that extend beyond the GMUG boundary or that have developments on 

private inholdings. 

Road and Trail Stream Crossings - number of open road and trail crossings per miles of perennial and 

intermittent streams within a given subwatershed. Stream crossings were determined by intersecting 

perennial and intermittent streams with existing and open roads and trails. Figure 5 shows where these 

stream crossings occur. The crossing count for a given subwatershed was then divided by the miles of 

perennial and intermittent streams for that same subwatershed, to get a count of crossings per mile of 

perennial and intermittent streams within a given subwatershed. Counts of crossings per mile of perennial 

and intermittent streams by subwatershed ranged from 0 to 1.2. 

Note: The NHD Flowline and the roads and trails layers are approximations and the resulting intersection 

point layer is not necessarily an accurate representation of all actual crossings. This information also does 

not identify if the crossing is a culvert, a bridge, or a ford. Existing culvert and bridge inventories are not 

complete. These are data gaps that need to be addressed in the future. 




Figure 5. Road and Trail Stream Crossings 

Roads and Trails within Riparian Areas - miles of open roads and trails per square mile of riparian 

areas within a given subwatershed. This was determined by identifying those segments of open roads and 

trails that occur within riparian areas, for each subwatershed. This length was then divided by the square 

miles of riparian areas for each subwatershed. Figure 6 shows where roads and trails occur within riparian 

areas. Miles of open roads and trails per square mile of riparian areas within a given subwatershed ranged 

from 0 to 10. 




Figure 6. Roads and Trails in Riparian Areas 

Recreation Developments within Riparian Areas - density of recreation developments per square mile 

of riparian areas within a given subwatershed. This was determined by identifying where recreation 

developments (i.e., campgrounds, picnic grounds, trailheads, parking areas, toilets) occur in riparian areas 

and dividing the count of these occurrences by the square miles of riparian areas for each subwatershed. 

Note: Only developed recreation sites were included; dispersed sites without structures were not. Figure 7 

shows where recreation developments occur within riparian areas. Recreation developments within 

riparian areas occur in 28 subwatersheds. Densities within riparian areas range from less than one to nine. 

Recreation Residences within Riparian Areas - density of recreation residences per square mile of 

riparian areas within a given subwatershed. This was determined by identifying where recreation 

residences occur within riparian areas. Note: Only those recreation residences that are permitted were 

included; residences that occur on private inholdings or areas outside the Forest boundary were not. 

Figure 7 also shows where recreation residences occur within riparian areas. Recreation residences within 

riparian areas occur in two subwatersheds. Densities ranged from less than one to three. 




Figure 7. Recreation Residences and Developed Sites in Riparian Areas 

Initially, water use diversions and storage structures were going to be included as part of the infrastructure 

value. These values were ultimately considered in the Water Use Values, below, where both the structures 

and the amount of the associated water use (in acre feet for storage or cubic feet per second for flow) were 

evaluated. 

It is not enough to know which subwatersheds have the most infrastructure values. Two different 

watersheds may have the same number of road and trail stream crossings, but there may be twice as many 

miles of streams in one watershed than the other, which could potentially have much larger stream flows 

and sediment/debris loads that could impact the crossings. Metrics used were designed to compare 

subwatersheds in a more relative way. For each individual infrastructure value, the results were 

standardized (results for each subwatershed were divided by the largest result of all the subwatersheds). 

The standardized results for each infrastructure value were then summed to get a cumulative 

infrastructure value (Stream Crossings + Roads and Trails in Riparian Areas + Recreation Developments 

in Riparian Areas + Recreation Residences in Riparian Areas = Infrastructure Value Ranking). The 

cumulative Infrastructure Value Rankings were classified into quartiles. The top 25% were classified 3 

(high), middle 50% were classified 2 (moderate), lowest 25% were classified 1 (low). Figure 8 shows the 

resulting Infrastructure Values Ranking. 




Figure 8. Infrastructure Values Ranking 

The Upper Taylor geographic area has the largest area in a high ranking for infrastructure values. The 

Cochetopa geographic area has the second largest area in high infrastructure values ranking, due primarily 

to road and trail stream crossings and miles of routes within riparian areas. The San Juans geographic area 

has the third highest amount of area in a high ranking for infrastructure values, mostly due to the density 

of road and trail stream crossings, with a few subwatersheds having a higher density of developed sites 

within riparian areas. The Uncompahgre geographic area has the lowest ranking for infrastructure values. 

Water Uses Values 

An initial purpose of the National Forest system was and remains to “secure favorable conditions of water 

flows.” Many water use values depend upon the runoff generated from the GMUG. Those values are 

realized both on and off the Forest. Water use values are both consumptive and non-consumptive. For this 

WVA, both public and private water uses were evaluated, and are listed below. 

Water Rights Quantification - acre feet per acre of subwatershed for water storage rights, or cubic feet 

per second per acre of subwatershed for water flow rights. Water rights included were those held by the 

US Forest Service, municipalities and other public entities, as well as private individuals and water user 

groups. Water uses associated with these rights are primarily for irrigation and stockwater, with some 

domestic water use. Data used to identify water rights originated with the State of Colorado Division of 

Water Resources. The state’s Division 4 overlaps all but the northern half of the Grand Mesa on the 

GMUG, which is within the State’s Division 5. Data for Division 4 included water rights/uses both on and 

off National Forest system land; Division 5 data used in this analysis were only for National Forest 

system land on the GMUG. Only actual, developed water rights were included. Water rights exist for 

approximately 1,704,070 acre feet of storage (quantification of water rights in acre feet per acre of 

subwatershed ranged from 0 to 79) and 24,620 cubic feet per second flow (quantification of water rights 




in cubic feet per second per acre of subwatershed ranged from 0 to 0.3). Figure 9 shows approximate 

locations where water rights occur in the WVA analysis area. NOTE: Many water rights locations in the 

state’s data are based on approximate quarter quad descriptions and not actual coordinates. 

Water Rights Structures - count of structures associated with each quantified water right per acre of 

subwatershed. The state’s data identify the type of structures associated with each water right. This varies 

among ditch, well, reservoir, pipeline, spring box and pump. There are 9,775 structures associated with 

water rights, with counts per acre of subwatershed ranging from 0 to 0.01.The water rights locations in 

Figure 9 are the approximate locations of these structures. 

Figure 9. Water Rights 

Surface Source Water Protection Areas - percent of source water protection area on GMUG by 

watershed. A number of communities rely on surface and groundwater originating on the GMUG NFs for 

their public drinking water supplies. Analysis of surface community water supplies previously conducted 

for the Forest plan revision process was used for this WVA. This analysis was limited to lands within the 

GMUG. There are a total of 18 surface water providers (32 separate systems or source water areas) that 

include at least some GMUG-administered lands. These source areas include portions of one or more 

subwatersheds on the GMUG. The source areas range from 500 acres to over 2 million acres in size, with 

the proportion lying within GMUG NFs varying from approximately 4% to 100%. Generally, the greater 

the proportion of NF lands in a source water area, the greater the potential to be directly affected by 

Forest Service land use and management activities. GMUG lands are considered the principal source of 

water where 70% or more of the total supply area lies within the Forest boundary. Forest-wide that 

includes 21 separate systems (managed by 16 providers), totaling approximately 1,038,000 acres. Figure 

10 shows subwatersheds where greater than 70% of a given source water protection area is on the Forest 

in pink. Portions of the GMUG that are included in source water areas where less than 70% is on the 

Forest are shown in green. 




Figure 10. Source Water Areas 

Instream Flow Water Rights - miles of instream flow water rights per square mile of subwatershed. 

The Colorado Water Conservation Board (CWCB) holds instream flow water rights on approximately 

1,800 miles of stream in 107 subwatersheds across the Forest (Figure 11). The quantity and timing of 

those flows vary by individual stream, but the CWCB program objective is to “preserve and improve the 

natural environment to a reasonable degree.” This nonconsumptive water use is designed to retain a 

minimum amount of water within a given stream, to protect the natural environment (which can include 

coldwater fisheries and riparian habitats, among other environmental factors). 




Figure 11. Instream Flow Water Rights 

Other water uses, such as water dependent recreation (fishing, rafting, kayaking), were initially 

considered but eventually eliminated from the WVA because they were either limited in their distribution 

or were represented by other values (e.g., fishing would occur where cold water fisheries are present). 

As with the infrastructure values, above, water use value metrics were designed to compare 

subwatersheds in a more relative way. For each individual water use value, the results were standardized 

(results for each subwatershed were divided by the largest result of all the subwatersheds). The 

standardized results for each water use value were then summed to get a cumulative water use value 

(water rights quantification + water rights structure + Surface Source Water Protection Areas + instream 

flow water rights = Water Uses Value Ranking). The cumulative Water Uses Value Rankings were 

classified into quartiles. The top 25% were classified 3 (high), middle 50% were classified 2 (moderate), 

lowest 25% were classified 1 (low). Figure 12 shows the resulting Water Uses Values Ranking. 




Figure 12. Water Uses Values Ranking 

The Upper Taylor geographic area has the largest area ranked high for water uses. The San Juans 

geographic area has the second-highest area ranked high for water uses. Because the water rights 

information from the State of Colorado did not include off-Forest data for Division 5 (the northern half of 

the Grand Mesa), the rankings for these subwatersheds are lower than they should be. Once data are 

acquired, the rankings should be re-evaluated for these subwatersheds. The Grand Mesa geographic area 

has several subwatersheds with the highest rankings for water use values on the GMUG. The 

Uncompahgre geographic area has the least area ranked high for water uses. 

Aquatic Ecological Values 

Aquatic Ecological Values identified for this WVA include both habitats and species. The GMUG team 

focused on those aquatic habitats and species that were of most concern and that would be representative 

of other aquatic habitats/species not selected. As with the other values, a mixture of data extent and 

availability for different aquatic values affects the confidence in the resulting watershed rankings. The 

aquatic ecological values included in this WVA are listed below. 

Fens, wetlands and riparian areas - density of riparian habitats measured as acres of habitat per square 

mile of subwatershed. A combination of a recent fen/wetland inventory database and an existing riparian 

habitat layer were used to identify where these aquatic habitats occur on the GMUG. Densities ranged 

from 0 to 121 acres of riparian habitats per square mile of subwatershed. (Data were limited to lands 

within the GMUG boundary.) The existing riparian habitat layer also includes waterbodies, so 

waterbodies were not evaluated separately. Figure 13 displays fens, wetlands, and riparian areas. 

Waterbodies are also display in this figure. 




Figure 13. Fens, Wetlands, Riparian Areas, and Waterbodies 

Coldwater Fisheries - miles of third order or higher perennial streams compared to the miles of perennial 

and intermittent streams in a subwatershed. An inventory of existing coldwater fisheries does not exist for 

the GMUG. We assumed that third order or higher perennial streams (not including crenulations) were 

likely to support salmonid fishes (brook trout (Salvelinus fontinalis), brown trout (Salmo trutta), rainbow 

trout (Oncorhynchus mykiss), and cutthroat trout (O. clarkii)) and associated fisheries. There are 

approximately 2,300 miles of third order or higher perennial streams identified on the GMUG. Figure 14 

displays these streams. (Note: not all perennial streams on the GMUG are considered to be third order or 

higher, so some fisheries habitat may have been overlooked in this evaluation. Lake and reservoir 

fisheries were not included because an inventory is lacking.) 

Cutthroat Trout Fisheries - miles of streams occupied by cutthroat trout per miles of coldwater fisheries 

streams by subwatershed. Native cutthroat trout populations on the GMUG include both the Colorado 

River and greenback lineages of Colorado River cutthroat trout (O. c. pleuriticus). Known occurrences of 

conservation populations of native cutthroat trout were included in this analysis. Conservation 

populations are those having less than 10 % non-native genes (Hirsch et al. 2006). These populations 

represent the highest conservation priority for fisheries resources on the GMUG. Figure 14 shows the 

extent of known cutthroat trout conservation populations. 




Figure 14. Coldwater Fisheries and Known Cutthroat Trout Occurrences 

Initially, the list of aquatic ecological values to be evaluated in this WVA was more extensive. Springs 

were identified as an important resource value likely to be affected by climate change; however, the 

spring inventory for the Forest is very limited. Boreal toad (Anaxyrus boreas boreas, a sensitive species) 

was not included because known occurrences are limited to very few sites on the Forest, and evaluation of 

effects to riparian habitats would address the effects to boreal toads and other amphibian species. Four 

warm water-sensitive fish species (bluehead sucker (Catostomus discobolus), flannelmouth sucker (C. 

latipinnis), mountain sucker (C. platyrhynchus), and roundtail chub (Gila robusta)) were also not 

included in the WVA because of limited data on occurrence and stream temperatures. Botanical species 

and communities were eliminated from consideration because general effects to their habitat would also 

be addressed through riparian habitats. 

Aquatic ecological value metrics were designed to compare subwatersheds in a more relative way. For 

each individual value, the results were standardized (results for each subwatershed were divided by the 

largest result of all the subwatersheds). The standardized results for each value were then summed to get a 

cumulative aquatic ecological value (Fen/wetland/riparian habitat + coldwater fisheries + cutthroat trout 

fisheries = Aquatic Ecological Value Ranking). The cumulative Aquatic Ecological Value Rankings were 

classified into quartiles. The top 25% were classified 3 (high), middle 50% were classified 2 (moderate), 

lowest 25% were classified 1 (low). Figure 15 shows the resulting Aquatic Ecological Values Ranking. 




Figure 15. Aquatic Ecological Value Rankings 

The Grand Mesa has the largest area with high rankings for aquatic ecological values, primarily due to the 

dense concentration of riparian and wetland areas and associated waterbodies. The Grand Mesa also has 

subwatersheds with cutthroat trout populations. The Upper Taylor geographic area has the second largest 

area with high rankings, also for a combination of aquatic habitats as well as cutthroat trout populations. 

The lower, drier Uncompahgre geographic area has the lowest rankings for aquatic ecological values. 

EXPOSURE 

Information on exposure, or the predicted climate changes anticipated to occur on the GMUG, came from 

a variety of sources. Published climate change reports for the State of Colorado were used as sources for 

predicted climate changes (Colorado Water Conservation Board Draft 2010; Ray et al. 2008; Spears et al. 

2009). This information was downscaled from global circulation models to the State of Colorado and the 

Upper Colorado River Basin. Further downscaled information was obtained from a report describing 

several climate and hydrologic change scenarios for the Upper Gunnison River (Barsugli and Mearns 

Draft 2010). Regional implications of climate change to fisheries information came from Rieman and 

Isaak (2010). Data modeled using the Variable Infiltration Capacity (VIC) hydrologic model were also 

used to evaluate potential climate changes for the GMUG. These data are described below. 

Anticipated Climate Change 

State of Colorado 




Climate change projections for the State of Colorado are summarized in “Climate Change in Colorado: A 

Synthesis to Support Water Resources Management and Adaptation” (Ray et al. 2008) and include the 

following projections. 

1. In Colorado, temperatures have increased about 2 ˚F in the past 30 years. Climate models project 

Colorado will continue to warm 2.5 ˚F [+1.5 to +3.5 ˚F] by 2025, relative to the 1950-99 baseline, 

and 4 ˚F [+2.5 to +5.5 ˚F] by 2050. The 2050 projections show summers warming by +5 ˚F [+3 to 

+7 ˚F], and winters by +3 ˚F [+2 to +5 ˚F]. 

2. Winter projections show fewer extreme cold months, more extreme warm months, and more 

strings of consecutive warm winters. 

3. In all seasons, the climate of the mountains is projected to migrate upward in elevation, and the 

climate of the desert southwest is projected to progress up into the valleys of the Western Slope. 

4. Variability in annual precipitation is high and no long-term trend in annual precipitation has been 

detected for Colorado. Multi-model average projections show little change in future annual mean 

precipitation, although seasonal shift in precipitation does emerge. 

5. Dramatic declines in lower-elevation (< 8,200 ft) snowpack are projected, due to more winter 

precipitation coming as rain than snow. Modest declines in snowpack are projected (10%-20%) 

for Colorado’s high-elevations (> 8,200 ft) by 2050. 

6. Between 1978 and 2004, the onset of spring runoff from melting snow has shifted earlier by two 

weeks. By 2050, the timing of runoff is projected to shift earlier in the spring, and late-summer 

flows may be reduced. These changes are projected to occur regardless of changes in 


7. The Upper Colorado River Basin average runoff is projected to decrease as much as 20% by 

2050, compared to the 20 th century average. 

8. Increased storm intensity and variability are projected to elevate risks for floods and droughts. 

9. Increasing temperature and soil moisture changes may shift mountain habitats higher in elevation. 

Forest, rangeland, and riparian plant communities may change with more xeric, drought-tolerant 

species becoming more abundant. 

10. More extensive wildfire activity, especially at lower elevation/fire dominated ecosystems is 

predicted. 

11. Decreased snowpack and earlier spring melt could diminish recharge of subsurface aquifers that 

support late summer and winter baseflows. 

Downscaled Scenarios for Gunnison Basin for 2040-2060 

Downscaled climate changes were also available for the GMUG. Barsugli and Mearns (Draft 2010) 

developed two climate change scenarios for a Climate Change Adaptation Workshop for Natural 

Resource Managers in the Gunnison Basin, facilitated by The Nature Conservancy. These scenarios were 

specifically designed to represent a “moderate” and a “more extreme” scenario for the 2040-2060 

timeframe. These scenarios were designed using the A2 emissions scenario because the world is already 

on this scenario path. Two hydrologic change scenarios were developed in tandem with the climate 

change scenarios, which produced quantitative estimates of how soil moisture, snowpack, and runoff 

would change, consistent with the temperature and precipitation change scenarios. These hydrologic 

scenarios were developed using the Sacramento Soil Moisture Accounting hydrology model, coupled to 

the “Snow-17” snow model, developed by the NOAA. 




These two scenarios describe a range in climate change predictions that may occur on the GMUG. The 

predictions are consistent with the state-wide changes described above, and further refine the potential 

effects that may be seen on the GMUG. 

Table 1 displays the predicted annual and seasonal changes in precipitation and temperature for the 

“moderate” scenario. 

Season Precipitation (%) Temperature (˚C) Temperature (˚F) 

Annual ~0.0 +2.0 to +3.0 +3.6 to +5.4 

Winter +15.0 +2.0 +3.6 

Spring -12.0 +2.5 +4.5 

Summer -15.0 +3.0 +5.4 

Fall +4.0 +2.5 +4.5 

Table 1. Temperature and Precipitation Changes for “Moderate” Climate Change Scenario developed 

by Barsugli and Mearns for the Gunnison Basin 

Predicted changes under the “moderate” scenario include: 

1. Increase in annual temperatures of 2-3 ˚C (3.6-5.4 ˚F). 

2. No substantial change in annual precipitation, but an increase in cool season precipitation and a 

decrease in warm season precipitation. 

3. Decrease in annual natural stream flows of 5% to 10%, due to increased temperature, even if 

annual precipitation remains the same. 

4. Warming temperatures lead to a later accumulation of snow in the fall and earlier snowmelt in the 

spring. However, because of the increased precipitation in winter and the generally cold, highelevation 

nature of the upper Gunnison basin, the mid-winter snowpack may be similar to the 

present. 

5. Snowmelt-driven stream flow will occur earlier in the spring by about a week on average. (Note: 

this shift is due to warming and does not include the effects of dust-on-snow, which can result in 

an even earlier shift in snowmelt.) 

6. The earlier melt, along with decreased summertime precipitation and increased summertime 

temperatures, results in lower amounts of water stored in the soils during summer and fall. 

Table 2 displays the predicted annual and seasonal changes for the “more extreme” scenario. The “more 

extreme” scenario is warmer and drier than the “moderate” scenario. 




Season Precipitation (%) Temperature (˚C) Temperature (˚F) 

Annual -10.0 +3.0 +5.4 

Winter ~0.0 +3.0 +5.4 

Spring -15.0 +3.0 +5.4 

Summer -20.0 +4.0 +7.0 

Fall -10.0 +3.0 +5.4 

Table 2. Temperature and Precipitation Changes for “More Extreme” Climate Change Scenario 

developed by Barsugli and Mearns for the Gunnison Basin 

Predicted changes under the “more extreme” scenario include: 

1. Increase in annual temperatures of 3 ˚C (5.4 ˚F). 

2. A 10% decrease in annual precipitation, with greater decreases in warm season precipitation. 

3. Decrease in precipitation and increase in temperature, both act to reduce annual stream flow totals 

in the range of 20% to 25%. 

4. Warming temperatures lead to a later accumulation of snow in the fall and earlier snowmelt in the 

spring. Because this likely represents a hot/dry scenario for much of the west, the potential exists 

for more frequent dust deposition events, which also may lead to an earlier melt and to reduced 

water yield from the snowpack. 

5. Snowmelt-driven stream flow will peak about two or more weeks earlier in the spring, though this 

effect may be less if dust effects on snowmelt are strong. The combined effects of dust and 

temperature on snowmelt timing tend to be dominated by the dust effects. 

6. The much earlier melt, along with decreased summer precipitation and increased summer 

temperatures, will result in extremely low amounts of water stored in the soils during summer and 

fall. 

VIC Model Climate Change Predictions 

The primary predictive model used to display climate changes was the VIC hydrologic model. Data 

derived using the VIC model were available from the Climate Impacts Group (CIG) at the University of 

Washington. Historic trends were developed from the climate record from 1916 to 2006. Future 

prediction results for temperature- and precipitation-related parameters were generated using: 1) a 

composite of the 10 climate models that best resembled the historic trend, 2) the MIROC_3.2 model 

(more extreme temperature increases), and 3) the PCM1 model (less extreme temperatures increases) for 

two time periods (2030-2059 and 2070-2099) using the A1B emissions scenario. Data were available at 

the ~6 km-grid scale for monthly averages for 21 parameters for each model, but not all parameters were 

reviewed by the GMUG team. (Data downloaded from 

ftp://ftp.hydro.washington.edu/pub/climate/USFS_monthly_summaries/CO/ on 11/5/2010). 

In addition, some of the data were summarized at the HUC-5 scale. (Data downloaded from 

ftp://ftp2.fs.fed.us/incoming/gis/PNF/WVA/ on 10/22/2010). Outputs obtained from the VIC Model data 

are described below. 




Initially, we reviewed the HUC-5 data for the composite, and MIROC_3.2 models, comparing projections 

of historic condition with two time periods (2030-2059 and 2070-2099) for the following parameters: 

• precipitation (monthly total, seasonal* total) 

• tmax (daily maximum temperature monthly average, seasonal* average) 

• tmin (daily minimum temperature monthly average, seasonal* average) 

• runoff (monthly total, seasonal* total) 

• baseflow (monthly total, seasonal* total) 

• hydrograph (runoff + baseflow as monthly total, seasonal* total) 

*Seasonal breakdown: winter = December, January, February; spring = March, April, May; 

summer = June, July, August; fall = September, October, November 

Charts for each HUC-5 were created to compare the composite and MIROC_3.2 model results to the 

historic trend for these parameters (this information is available as GMUG Appendix A at 

www.fs.fed.us/ccrc/wva/appendixes). (Note: We did not chart the PCM1model results that averaged 

between the composite and MIROC_3.2 results). For most HUC-5 watersheds, the data display future 

decreases in summer and fall precipitation and shifts in precipitation between winter and spring. 

Temperature increases of 2 to 3 ˚C are predicted for both maximum and minimum temperatures 

throughout the year. Runoff periods are predicted to shift one to two months earlier and total runoff is 

reduced. (Note: these predictions are in addition to the changes already seen since 1978, described 

earlier.) 

Because some HUC-5 watersheds include a wide range of elevations (ranges of 5,000 to 7,000 feet), we 

also reviewed the 6 km-grid scale VIC data. Predicted results for the composite and MIROC_3.2 models 

were compared to the historic trend for the same parameters listed above, as well as for 

evapotranspiration. We looked at the actual change between modeled and historic results, and the percent 

change on a monthly basis at the 6 km-grid scale. Maps showing monthly results at the grid scale display 

large differences between higher and lower elevation areas (see this information is available as GMUG 

Appendix B at www.fs.fed.us/ccrc/wva/appendixes). 

We used the six geographic areas (areas with similar climatic regimes and elevation ranges) to examine 

predicted climate changes (see Figure 4). Since most of the lower elevations within the HUC-5 scale 

watersheds are actually below the GMUG Forest boundary, reviewing exposure parameters at the 

geographic area scale is more representative for the GMUG. 

We chose to focus on a smaller subset of VIC parameters at the geographic area scale. We compared the 

predicted seasonal temperature changes (both maximum and minimum averages) from the MIROC_3.2 

model to the historic model. Figure 16 displays the seasonal increase in maximum average temperature by 

geographic area. Figure 17 displays the seasonal increase in minimum average temperature by geographic 

area. 




°C 

Figure 16. Seasonal Increase in Maximum Average Temperature by Geographic Area 

°C 

4 

3.5 

3 

2.5 

2 

1.5 

1 

0.5 

0 

4 

3.5 

3 

2.5 

2 

1.5 

1 

0.5 

0 

Seasonal Increase in Maximum Average Temperature 

by Geographic Area 

Spring Summer Fall Winter 

Seasonal Increase in Minimum Average Temperature 

by Geographic Area 

Spring Summer Fall Winter 

Figure 17. Seasonal Increase in Minimum Average Temperature by Geographic Area 

Temperatures are predicted to increase across all seasons and across all geographic areas. Increases in 

minimum daily temperatures will be very similar to increases in maximum daily temperature. Spring 

temperatures are expected to increase the most for the Uncompahgre Plateau, San Juans, Grand Mesa, and 

West Elk geographic areas. For the Uncompahgre Plateau, this spring increase may mean the difference 

from being below freezing to above freezing, which will change the precipitation from snow to rain, and 

which could affect snowpack melt and stream flow response. Summer temperatures are expected to 

increase the most for the more easterly geographic areas (Upper Taylor and Cochetopa). Fall temperatures 

are expected to increase the least for all geographic areas. However, for the Uncompahgre Plateau and the 

Grand Mesa, this increase could extend the frost-free period, resulting in longer growing seasons and later 


Uncompahgre 

San Juans 

Cochetopa 

Upper Taylor 

West Elk 

Grand Mesa 

Uncompahgre 

San Juans 

Cochetopa 

Upper Taylor 

West Elk 

Grand Mesa



onset of snowpack. The largest annual increase in temperatures is predicted for the Uncompahgre Plateau, 

followed in order by Grand Mesa, San Juans, West Elk, Upper Taylor, and Cochetopa. 

An aridity index was used to forecast where water availability may be most affected. By determining the 

ratio of precipitation to potential evapotranspiration, we identified, in a very simplistic way, those 

locations where water surpluses or deficits are most likely to occur. A reduction in precipitation with an 

increase in potential evapotranspiration will reduce soil moisture, fuel moisture, groundwater recharge, 

and availability of water to contribute to sustained stream flow. An aridity index of 1.0 means 

precipitation meets the demand of potential evapotranspiration. An aridity index of less than 1.0 means 

potential evapotranspiration exceeds precipitation and plants are under water stress. An aridity index 

greater than 1.0 means precipitation exceeds potential evapotranspiration and there is available water in 

the system. We compared the change in the seasonal aridity index for the MIROC_3.2 model to the 

historic trend (Figure 18). 

Aridity Index 

Aridity Index 

3.5 

3.0 

2.5 

2.0 

1.5 

1.0 

0.5 

0.0 

70 

60 

50 

40 

30 

20 

10 

0 

Historic 

MIROCPrediction 

Historic 


Spring 

Winter 

Figure 18. Seasonal Aridity Indices by Geographic Area 

The MIROC_3.2 model predictions indicate a significant change in aridity indices throughout the year, 

but once again, spring appears to be the season that may be most affected by climate change. Historically, 

only the Uncompahgre Plateau has had an aridity index below 1.0 in the spring. Predictions from the 

MIROC_3.2 model indicate the Cochetopa and West Elk geographic areas may also become waterstressed 

in the spring. All geographic areas have had and will continue to have aridity indices below 1.0 in 

the summer. Water availability has not generally been a limiting factor in the fall for any of the 

geographic areas, but the aridity index is expected to drop to less than 1.0 for the three driest geographic 


Aridity Index 

3.5 

3.0 

2.5 

2.0 

1.5 

1.0 

0.5 

0.0 

Aridity Index 

3.5 

3.0 

2.5 

2.0 

1.5 

1.0 

0.5 

0.0 

Historic 


Historic 


Summer 

Fall



areas (Cochetopa, Uncompahgre, and West Elk). The amount of available water is expected to become 

limiting in the Uncompahgre, Cochetopa, and West Elk geographic areas for three out of four seasons. 

Figure 19 displays the annual change in aridity indices for both the composite and MIROC_3.2 models, 

compared to the historic trend. All geographic areas are predicted to become drier. The largest changes 

will actually occur at the highest elevations (San Juans, Upper Taylor, and Grand Mesa) in those 

geographic areas with the highest precipitation. These areas also have the greatest capacity to buffer the 

effects of climate change because of the high levels of water produced from snowmelt and higher 

occurrence of aquatic habitats. These areas also support high levels of water development for human uses, 

so any increase in aridity could have a dramatic effect on water uses. This is potentially a very big 

concern in the Grand Mesa geographic area, where the annual aridity index is predicted to drop below 1. 

Airidity Index (mm/mm) 

5.0 

4.5 

4.0 

3.5 

3.0 

2.5 

2.0 

1.5 

1.0 

0.5 

0.0 

-‐11.18% 

-‐23.35% 

Historic 

COMP 

MIROC 

-‐9.37% 

1.29 

-‐18.31% 

-‐11.87% 

1.07 

-‐22.00% 

Figure 19. Annual Change in Aridity Index by Geographic Area 

0.88 0.87 

Geographic areas were ranked for exposure, based on the predicted changes (from the MIROC_3.2 

model outputs) for maximum and minimum temperatures and the annual percent change in aridity index 

(Table 3). A score of 1 indicates lower exposure; a score of 6 indicates higher exposure. Figure 20 

displays this ranking. 


-‐8.72% -‐9.53% -‐8.57% 

-‐18.05% -‐18.79% -‐18.80% 

San Juans Upper Taylor Grand Mesa West Elk Cochetopa Uncompahgre 

0.75 

0% 

-‐10% 

-‐20% 

-‐30% 

-‐40% 

-‐50% 

-‐60% 

-‐70% 

-‐80% 

-‐90% 

-‐100% 

Percent Change (%)



Geographic Area 

Tmin 

Rank* 

TMax 

Rank* 

Aridity 

Index Rank* 


Exposure 

Rank** 

Exposure 

Rank 

(numeric) 

Uncompahgre 6 6 4 0.89 6 

Grand Mesa 5 5 5 0.83 5 

San Juans 4 4 6 0.78 4 

West Elk 3 3 1 0.39 3 

Upper Taylor 2 2 2 0.33 2 

Cochetopa 1 1 3 0.28 1 

Table 3. Geographic Area Exposure Ranking 

* Highest number has most change 

** (Tmin Rank + Tmax Rank + Aridity Index Rank) / 18 

Figure 20. Geographic Area Exposure Ranking 

Table 4 summarizes key potential climate changes described above and their potential effects to 

hydrologic process and identified aquatic resource values. This table was modified from Table 2 

found in Furniss et al (2010).



Projected Climate Change Anticipated Hydrologic Response 

Warmer Winter/Spring Temperatures 

Average daily winter/spring temperature 

expected to increase > 3 ˚C by 2050. 

Warmer Summer Temperatures 

Average daily summer temperature 

expected to increase > 3 ˚C by 2050. 

Changes in Precipitation 

At higher elevations, may be slightly 

greater precipitation during the winter, 

but likely less total precipitation, 

especially during warmer months. 

• Fewer extreme cold months, 

more frequent extreme warm 

months, more consecutive 

warm winters 

• Later accumulation of 

snowpack. 

• Earlier onset of snowpack 

runoff (1-3 weeks) 

• Higher winter stream flows 

• Increased water temperature 

• Winter precipitation more 

often rain than snow below 

8200 feet 

• Snowline to move up in 

elevation. 

• Increased evapotranspiration 

• Decreased soil moisture 

• Reduced summer stream flows 

• Increased water temperature 

• May see higher peak flows 

associated with snowmelt, 

earlier in the year. 

• Lower summer and fall 

baseflows 

• Increased soil moisture during 

spring at lower elevations 



Resource Values 

• Reduced duration of winter 

snow cover 

• Longer period of saturated 

roadbeds vs. frozen roadbeds 

• Increased demand for water 

storage 

• Earlier demand for irrigation 

water 

• Decreased summer stream 

flows 

• Potential change to aquatic 

species reproductive triggers 

or success 

• Increased risk to channel and 

floodplain infrastructure 

from higher runoff 

• Increased risk to riparian 

habitat/floodplains from 

higher flows 

• Changes to winter habitat, 

winter recreation and plant 

communities 

• Increased demand for 

irrigation water 

• Shifts in cold water habitat to 

higher elevations 

• Increases in warm water 

habitat 

• Decreased dissolved oxygen 

in lower elevation streams 

during the summer 

• Aquatic biota mortality and 

even loss of populations 

• Loss of summer stream flow 

• Decreased water availability 

during irrigation season 



• Reduced riparian vegetation 

health and vigor 

• Increased landslides and 

slumps on geologically 

unstable areas 

• Increased potential damage 

to saturated roadbeds 

• Reduced aquatic habitat in 

summer and fall



Projected Climate Change Anticipated Hydrologic Response 

More intense storms 

Warmer atmosphere has potential for 

increase in frequency and magnitude of 

big storms. 

More frequent and longer periods of 

drought 

Increase winter dust deposition on 

snowpack 

• Localized flooding 

• Increased debris flows 

• Increased hillslope and channel 

erosion 

• Less soil moisture 

• Reduced groundwater recharge 

• Lower summer and fall 

baseflow 

• Accentuate changes to 

snowpack melt 



Resource Values 



from sediment and high 

flows 

• Increased concern for public 

safety 

• Increased selenium load in 

streams where Mancos Shale 

exposure is significant. 

• Increased erosion associated 

with natural disturbances 

associated with drought (e.g., 

fire) 

• Increased plant stress and 

susceptibility to insect and 

disease mortality 

• Reduced groundwater 

contribution to baseflows 

• Reduced discharge from 

springs 

• Reduced wetland/riparian 

function 

• Similar to warmer winter 

consequences 

Table 4. Projected climate changes to the GMUG NF, anticipated hydrologic response and potential consequences 

to aquatic resource values 

WATERSHED RISK 

Inherent characteristics and past management of watersheds influence how a watershed is likely to be 

affected by climate change, and when combined, can be considered as contributors to watershed risk. 

Some characteristics and/or impacts from past activities may exacerbate the anticipated impacts of 

climate change (stressors), while others may reduce the impacts of climate change (buffers). 

Inherent characteristics of watersheds were evaluated as two types of sensitivities on the GMUG: 1) 

sensitivity to erosion or sediment production, and 2) sensitivity to runoff response. Existing condition was 

evaluated based on past management activities. (The GMUG has not yet completed the new watershed 

condition classification, as directed by the Washington Office.) 

Sensitivities are described below. 

Erosion or Sediment Production Sensitivity 

The erosion or sediment production sensitivity was initially developed as part of the watershed 

assessment completed for the Forest plan revision. Characteristics of geology, soils, landforms and 

topography that affect the erosion potential or amount of sediment production from a given subwatershed 

were evaluated. Due to data limitations of some information, this evaluation was limited to lands within 

the GMUG Forest boundary. Mass wasting potential was not available at the time of the Forest plan



revision, but is currently available, and has been added to the suite of factors evaluated for this sensitivity. 

The list of factors evaluated for erosion or sediment production sensitivity include those listed below. 

1. Erosion Risk Rating – percent of severe and very severe erosion risk classes by subwatershed. 

This was derived from Kw factor (from soil survey data) and prevailing slope. The Kw factor is 

an indication of susceptibility of a soil to sheet and rill erosion by water, based on soil 

composition, structure, and permeability. The erosion risk rating was considered to be a stressor. 

2. Runoff potential – percent of subwatershed in Hydrologic Group D. Runoff potential is 

determined by soil infiltration capacity after prolonged wetting, permeability, depth to water 

table, and depth to restrictive or impervious layer. Soils with the highest potential for runoff are 

identified as Hydrologic Group D in soil survey data. Runoff potential was considered to be a 

stressor. 

3. Rainfall Intensity Factor – weighted average for each subwatershed. The rainfall intensity factor 

was derived from the Revised Universal Soil Loss Equation (RUSLE) R factor from PRISM data 

(obtained from Oregon State University). When other factors remain constant, soil loss is directly 

proportional to a rainfall factor related to the total quantity and intensity of rainfall. The RUSLE 

R factor is the average annual product of kinetic energy and maximum 30-minute rainfall 

intensity. The rainfall factor was considered to be a stressor. Based on the prediction that storm 

intensity is likely to increase, this factor is expected to increase in the future. 

4. Stream density – total miles of perennial and intermittent streams per square miles of 

subwatershed. This factor characterizes the degree of dissection and network transport capacity 

for both runoff and sediment. The higher the stream density, the larger the amount of sediment 

that may be moved through a subwatershed. Stream density was considered to be a stressor. 

5. Hydrologic Response Channels – percent of total stream network that is a response channel, 

compared to the total perennial and intermittent stream network in a subwatershed. Response 

channels are streams of third order or higher, with a gradient less than or equal to 1.5%, 

containing alluvial channel material, and classified as a Rosgen stream type of C, D or E. 

Response channels could be considered either buffers or stressors, depending on the situation. 

Response channels would be buffers in the situation where sediment is deposited in these areas 

and prevented from moving downstream. Response channels could also be added stressors 

because of the sediment loads they may retain, which under intense storms with high runoff could 

be released to impact downstream locations. 

6. Mass wasting potential – percent of a subwatershed with high mass wasting potential. Areas 

with mass wasting potential include areas with identified geological instability and areas with 

potential for mass wasting based on presence of vulnerable sedimentary geology and slopes 

greater than 50 percent. This factor was considered a stressor. 

Values for each of the individual factors listed above were calculated and then standardized for each 

factor (as described above for the values). The overall erosion or sediment potential sensitivity ranking 

was determined by adding the individual factor standardized ratings together for each subwatershed. The 

resulting Erosion Sensitivity Rankings were classified into quartiles. The top 25% were classified 3 

(high), middle 50% were classified 2 (moderate), and lowest 25% were classified 1 (low). Figure 21 

shows the resulting Erosion Sensitivity Ranking. 




Figure 21. Erosion Sensitivity Ranking 

Runoff Response Sensitivity 

The runoff response sensitivity was identified to show the relative ability of a subwatershed to produce 

rapid runoff following a storm event. This sensitivity is also based on inherent characteristics of the 

geology, soils, and basin characteristics (topography) of a watershed. Many of the factors included in this 

sensitivity are the same as those included in the erosion sensitivity described above, and the extent of this 

data was limited to lands within the GMUG boundary. Basin characteristics were calculated for entire 

subwatersheds both on and off the Forest. Factors that contribute to the flashiness of a given 

subwatershed include: 

1. Time of Concentration, a function of basin length (defined as the greatest distance from the 

watershed pour point to a point on the watershed divide which roughly follows the main 

drainage) and basin relief (the difference in elevation between basin pour point and highest point 

on the watershed boundary). Time of Concentration was considered to be a stressor. 

2. Stream Density – total miles of perennial and intermittent streams per square miles of 

subwatershed. This factor characterizes the degree of dissection and network transport capacity 

for both runoff and sediment. The higher the stream density, the larger the amount of runoff that 

may be moved through a subwatershed. Stream density was considered to be a stressor. 

3. Basin Ruggedness, a function of drainage density, basin relief and basin area. 

4. Rainfall Intensity Factor – weighted average for each subwatershed. The rainfall intensity factor 

was derived from the RUSLE R factor from PRISM data (obtained from Oregon State 

University). When other factors remain constant, soil loss is directly proportional to a rainfall 




factor related to the total quantity and intensity of rainfall. The RUSLE R factor is the average 

annual product of kinetic energy and maximum 30-minute rainfall intensity. The rainfall factor 

was considered to be a stressor. Based on the prediction that storm intensity is likely to increase, 

this factor is expected to increase in the future. 

5. Runoff potential – percent of subwatershed in Hydrologic Group D. Runoff potential is 

determined by soil infiltration capacity after prolonged wetting, permeability, depth to water 

table, and depth to restrictive or impervious layer. Soils with the highest potential for runoff are 

identified as Hydrologic Group D in soil survey data. Runoff potential was considered to be a 

stressor. 

6. Waterbodies, riparian and wetland areas – density of these aquatic features within a given 

subwatershed. Waterbodies and riparian and wetland areas were considered buffers to runoff 

response and the ratings for this factor were given negative values so they would buffer the 

combined runoff response ranking. 

7. Average annual baseflow – weighted average annual baseflow for each subwatershed. This 

value was determined from VIC data (modeled data for historic baseflow at the 6 km-grid scale). 

Baseflow is considered a buffer to runoff response and the ratings for this factor were given 

negative values so they would buffer the combined runoff response ranking. 

Values for each of the individual factors listed above were calculated and standardized for each factor (as 

described above for values). The overall runoff response sensitivity ranking was determined by adding the 

individual factor standardized ratings together for each subwatershed. The resulting Runoff Sensitivity 

Rankings were classified into quartiles. The top 25% were classified 3 (high), middle 50% were classified 

2 (moderate), and lowest 25% were classified 1 (low). Figure 22 shows the resulting Erosion Sensitivity 

Ranking. 




Figure 22. Runoff Sensitivity Ranking 

Past Management Activity Stressors 

Data used to evaluate past management or activity stressors were taken from the watershed assessment 

conducted for the Forest plan revision. (Additional discussion of the data uses, limitations of that data, 

and the effects of these anthropogenic stressors can be found in the Chapter 5, Section C of the watershed 

assessment completed for the Forest plan revision (2005).) A mix of long-term effects (e.g., dams and 

major roads) and short-term effects (e.g., timber harvests) have been included. Some stressors have direct 

effects on or near channels; others affect areas throughout a subwatershed. Several individual stressors 

were combined so that effects were not overweighed in the final subwatershed rankings. Data used for 

this evaluation were limited to areas within the GMUG boundary. For watersheds that have a large 

portion of off-Forest area, these rankings may need to be adjusted as off-Forest data become available. 

Individual activity stressors considered include those listed below. 

Flow Related Stressors 

1. Stream miles below diversions, expressed as a percentage of perennial and intermittent stream 

network in a watershed. There are some significant caps in understanding of the specific effects 

of diversions on aquatic systems. Operation information is only available for the major diversion, 

concerning timing and quantity of water diverted from or into the natural stream network 

2. Stream miles below reservoirs, expressed as a percentage of perennial and intermittent stream 

network in a watershed. Only reservoirs of 50 surface acres or larger were included. There are 

many smaller reservoirs and stockponds whose effects are not addressed; however, it was felt that 

these smaller reservoirs would have limited ability to influence flow regimes. Operation of larger 

reservoirs can regulate flows in ways that benefit fisheries and other aquatic values. 




3. Stream miles inundated by reservoirs, expressed as a percentage of perennial and intermittent 

stream network in a watershed inundated by reservoirs greater than 50 acres in size, because at 

that scale entire stream reaches or major wetland complexes would be impacted. 

Route Related 

1. Motorized route (roads and trails) density, expressed as miles of route per sq mi. of watershed. 

(Note: Travel management decisions made since 2005 are not reflected in these results.) 

2. Motorized route density within buffered riparian area, expressed as miles of routes within the 

area of riparian habitat and a 100-foot buffer around riparian habitat by watershed. 

3. Motorized route crossing density, expressed as number of crossing (determined by intersecting 

roads and trails layers with stream layer) compared to the total stream network (perennial and 

intermittent streams). 

Past vegetative treatments, expressed as a percentage of the watershed treated by some vegetation 

management within the past 50 years. 

High frequency of streamside recreational use, expressed as a percentage of the total miles of stream 

network in a watershed that have high levels of recreational use (camping, fishing, roads and trails, 

developed sites). 

Private land inholdings, expressed as a percentage of the total watershed area, was used as a measure of 

urban influences based on the assumption that as the amount of inholdings increases there is a greater 

potential for developments activities to be located on those private lands as opposed to surrounding NFS 

lands. 

Abandoned mine land site density, expressed as number of adits and tailings piles by area of each 

watershed. 

Values for each of the individual factors listed above were calculated and standardized for each factor (as 

described above for the values). The overall activity stressors ranking was determined by adding the 

individual factor standardized ratings together for each subwatershed. The resulting Activity Stressors 

Rankings were classified into quartiles. The top 25% were classified 3 (high), middle 50% were classified 

2 (moderate), and lowest 25% were classified 1 (low). Figure 23 shows the resulting Activity Stressors 

Ranking. 




Figure 23. Activity Stressors Ranking 

Method Used to Characterize Watershed Risk Due to Sensitivities and Stressors 

Watershed risk was evaluated in two ways, based on the two different sensitivities discussed above. Each 

sensitivity was combined with the activity stressors: 1) Erosion or Sediment Production Sensitivity 

combined with Activity Stressors, and 2) Runoff Response Sensitivity combined with Activity Stressors, 

with the resulting Watershed Risk rankings being determined using the following matrix. 




Sensitivity × Stressors Risk 

Ranking Matrix 

Sensitivity 

Sensitivity x Stressors 

Low Moderate High 

Low Low Low Low 

Moderate Low Low High 

High High High High 

The GMUG team working on the WVA felt that the inherent characteristics of a subwatershed would 

have greater influence on the overall watershed risk than the effects of past management activities. For 

this reason, if a subwatershed was ranked “High” for either one of the sensitivities, the watershed risk 

ranking was “High.” If the subwatershed ranking for either sensitivity was “Low,” the watershed risk 

ranking was “Low.” The following figures show the resulting watershed risk ranking for the erosion 

sensitivity combined with activity stressors (Figure 24) and the resulting watershed risk ranking for the 

runoff response sensitivity combined with activity stressors (Figure 25). In both figures, the 

subwatersheds with the highest risk are shown in red, and those with the lowest risk are shown in green. 

Figure 24. Erosion Sensitivity × Activity Stressors Ranking 




There are a total of 58 “High” risk subwatersheds for Erosion Sensitivity × Activity Stressors. The 

majority of these subwatersheds are found in the San Juans, Upper Taylor, and West Elk geographic 

areas. Twenty-three of these subwatersheds have a “High” Risk Rating just for Erosion Sensitivity × 

Activity Stressors alone, and 35 also have a “High” risk for Runoff Response Sensitivity × Activity 

Stressors (compare with Figure 25). 

Figure 25. Runoff Response Sensitivity × Activity Stressors Ranking 

There are 63 “High” risk subwatersheds for Runoff Response Sensitivity × Activity Stressors. The 

majority of these subwatersheds are found in the San Juans and Grand Mesa geographic areas. Of these, 

28 subwatersheds have a “High” risk rating for Runoff Response Sensitivity × Activity Stressors, while 

the remaining 35 are also “High” risk for Erosion Sensitivity × Activity Stressors (compare with Figure 

24). 

RESULTS (VULNERABILITY) 

To determine relative vulnerability of identified aquatic resources to predicted climate change, we need to 

combine all the pieces described above (resource values, risk [inherent sensitivity of the land and past 

management], and exposure) to see where they overlap. Resources of concern are most vulnerable where 

they occur in subwatersheds with highest sensitivity. The additional stress from climate change is most 

likely to have greatest impact in these areas. 

Method Used to Rank Resource Values Relative to Watershed Risk 

The different aquatic resource values of concern identified for this WVA can be affected by 

erosion/sedimentation and runoff in different ways. For this reason, the results of the two different risk 




rankings based on the different types of sensitivities were each related to the three aquatic resource 

values. The process used to compare the Resource Values Rankings to the Sensitivities x Stressors Risk 

Rankings is displayed in the following matrix. 

Values × Sensitivity Stressors 

Risk Ranking Matrix 

Values 

Sensitivity × Stressors 

Low High 

Low Low Low 

Moderate Low High 

High Low High 

Subwatersheds with a High Sensitivity × Stressor Risk Ranking and a High or Moderate Values ranking 

were rated as High. Subwatersheds with a High Sensitivity × Stressor Risk Ranking but a Low Values 

ranking were rated Low because of the reduced level of concern for the values. All Subwatersheds with a 

Low Sensitivity × Stressor Ranking were rated as Low when compared to Values because there is lower 

risk from the existing conditions within these subwatersheds. The results of the values related risk 

rankings are discussed below. 

Infrastructure Values Vulnerability 

Infrastructure in and near streams and rivers are vulnerable to flooding and/or sediment and debris flows 

that may result from climate change-related disturbances. These effects are most likely to occur in 

subwatersheds that have the highest risk due to inherent sensitivities for erosion or runoff response and a 

concentration of past management activities. 

Infrastructure values were related to Erosion Sensitivity × Activity Stressors with results displayed in 

Figure 26. Subwatersheds where infrastructure values are at the highest risk from erosion or sediment 

production are in the Upper Taylor, San Juans, and West Elk geographic areas. Infrastructure values were 

related to Runoff Response Sensitivity × Activity Stressors with results displayed in Figure 27. 

Subwatersheds with the highest risk from rapid runoff response are mostly in the San Juans, with some 

localized areas in the Grand Mesa, Upper Taylor, and Cochetopa geographic areas. 

Increased runoff could erode sections of roads and trails, and could wash out crossings and structures. 

High densities of roads and trails can collect overland flow and divert it into stream networks, adding to 

high flow conditions. Road networks with undersized pipes to accommodate existing flows will become 

more vulnerable. Increased sediment or debris loads could also plug culverts at crossings or bury sections 

of roads or structures. All results could threaten public safety and greatly increase maintenance costs. 




Figure 26. Risk Ranking for Infrastructure Values related to Erosion Sensitivities and Stressors 

Figure 27. Risk Ranking for Infrastructure Values related to Runoff Response Sensitivities and Stressors 




Warmer fall, winter, and spring temperatures can result in more precipitation falling as rain instead of 

snow, particularly at elevations less than 8,200 feet. Most of the GMUG NF is above 8,200 feet in 

elevation, so the chance of rain-on-snow related flood events is judged to be relatively minor. (Only the 

Uncompahgre geographic area has significant area at elevations below 8,200 feet.) Periods of freezing 

weather will likely be shortened, especially on the Uncompahgre and Grand Mesa geographic areas, and 

road and trail surfaces at lower elevations can remain saturated and subject to rutting for longer periods. 

Warmer winter and spring temperatures will also result in earlier and more rapid snowmelt runoff, which 

can result in flooding and increased sediment/debris flows. Dust-on-snow events have already been 

documented to result in earlier and more rapid snowmelt runoff, with or without temperature increases 

(Painter et al. 2010). 

The greater risk to infrastructure values has to do with an increased severity in summer thunderstorm 

events. Increased summer temperatures are likely to increase the potential energy associated with 

convective storm development. These types of storms can result in very high-intensity rainfall events, 

capable of localized flooding, and in certain geomorphic settings (i.e., those subwatersheds with high risk 

for erosion or sediment production), triggering debris flows that are capable of great damage and risk to 

life. While high intensity summer storms could potentially occur anywhere on the Forest, they historically 

occur most frequently in the San Juans geographic area. Considering all this information, infrastructure 

values are most vulnerable in the San Juans and Upper Taylor geographic areas. 

Water Use Values Vulnerability 

Water use values are vulnerable to predicted climate change impacts in several ways. Structures related to 

water use values (dams, reservoirs, ponds, ditches, diversions) are most vulnerable to flooding and/or 

sediment and debris flows, similar to infrastructure values. Water Use Values related to Erosion 

Sensitivity × Activity Stressors are shown in Figure 28. The areas where erosion or sediment potential has 

the highest risk of affecting water use values structures are highest in the Upper Taylor, San Juans, and 

West Elk geographic areas. Because off-Forest water use data were not available for the Grand Mesa 

geographic area, some additional subwatersheds on the Battlement and Sunnyside areas could actually 

have higher risk rankings related to erosion sensitivity. 

Water Use Values related to Runoff Response Sensitivity × Activity Stressors are shown in Figure 29. 

The areas where runoff potential has the highest risk of affecting water-use-value-related structures are 

mostly in the San Juans geographic area, with smaller groupings of subwatersheds in the remaining 

geographic areas. Increasing peak flow and duration of high-stage events could result in storage and/or 

diversion facilities being overtopped or washed away. Timing of runoff may also come at periods where 

storage structures are full or are normally releasing water in preparation for later seasonal inputs. 

Increased sediment loads that could result from flooding may fill in storage structures and diversions, 

reducing the amount of water these facilities could hold or transport; this could potentially increase 

maintenance costs to dredge, replace, or repair affected structures. Geographic areas where water use 

related structures are most vulnerable are the San Juans and Upper Taylor. 




Figure 28. Risk Ranking for Water Use Values related to Erosion Sensitivities and Stressors 

Figure 29. Risk Ranking for Water Use Values related to Runoff Response Sensitivities and Stressors 




Consumptive water use values (public and private water rights for irrigation, domestic and stock water 

use, and source water protection areas for communities) are vulnerable due to predicted changes in 

temperature and precipitation. Increased temperatures can alter the timing of runoff and lengthen the 

season of demand for water in the spring and fall. Aridity indices are expected to decrease even if 

precipitation does not change, because warmer temperatures will result in increased evapotranspiration. 

The result is potentially less available water for ecological processes and human use. Predicted reductions 

in annual precipitation, along with the potential for longer and more frequent droughts, further reduce 

water availability. Water will be most limited in those areas with aridity indices below 1.0 (Uncompahgre, 

West Elk, and Cochetopa). If these landscapes become more arid, existing water developments may no 

longer hold water, potentially reducing livestock management opportunities. Consumptive water uses on 

the Grand Mesa geographic area may be most vulnerable, because the aridity index is predicted to 

decrease to less than 1.0. It is not clear how the large concentration of existing waterbodies and associated 

riparian/wetland habitats found on the Grand Mesa may buffer predicted effects. 

Aquatic Ecological Values Vulnerability 

Similar to water use values discussed above, aquatic ecological values are vulnerable to predicted climate 

changes in several ways. Aquatic values, such as fisheries and riparian/wetland habitats associated with 

streams, are vulnerable to flooding and sediment/debris loading. Risk is exacerbated in subwatersheds 

that have inherent sensitivity and are impacted by past management activities. Aquatic Ecological Values 

related to Erosion Sensitivity × Activity Stressors are shown in Figure 30. The areas where erosion or 

sediment potential has the highest risk of affecting aquatic ecological values are highest in the Upper 

Taylor and San Juans geographic areas. Aquatic Ecological Values related to Runoff Response Sensitivity 

× Activity Stressors are shown in Figure 31. The areas where runoff potential has the highest risk of 

affecting aquatic ecological values are in the San Juans geographic area, with smaller groupings of 

subwatersheds in the remaining geographic areas. 

Flooding due to earlier and/or rapid runoff can result in scouring out of aquatic habitats, resulting in loss 

of vegetation and other habitat features, as well as flushing resident trout or eggs out of the most suitable 

habitats. Increased sediment loads could fill in aquatic habitats and riparian areas, as well as smother 

nesting gravels for stream-dwelling fish. Debris flows simplify channel habitats through removal of banks 

and large wood, especially in headwater streams with moderate to high gradient. Wetlands and offchannel 

habitats become filled with sediment, reducing the size and functionality of these habitats. 

Subwatersheds with aquatic ecological values in the Upper Taylor and San Juans geographic areas are 

most vulnerable to these combined effects. 




Figure 30. Risk Ranking for Aquatic Ecological Values related to Erosion Sensitivities and Stressors 

Figure 31. Risk Ranking for Aquatic Ecological Values related to Runoff Response Sensitivities 

and Stressors 




Fisheries and aquatic habitats can be directly affected by the predicted changes in temperature and 

precipitation. Temperature increase may have both negative and positive effects on cold-water fisheries in 

general and on cutthroat trout populations in particular. Occupied habitats at lower elevations may be 

eliminated as stream temperatures increase due to increases in air temperatures. The loss of cold-water 

fisheries may allow an expansion of occupied habitat for several sensitive species (e.g., bluehead sucker, 

roundtail chub) currently only found in streams and rivers at lower elevations. Increases in stream 

temperatures at higher elevations may actually benefit fish populations by making these streams more 

productive due to increasing growth rates of the fish that occupy them. Our current thinking is that low 

water temperatures in high-elevation streams limit fish growth and recruitment. Because current stream 

temperature data are lacking for most of the Forest, it is unknown if and specifically where low stream 

temperature could be having these effects. In 2011, the Forest began a multi-year project to collect and 

summarize baseline stream temperature data. The collection effort will focus on streams that support 

conservation populations of cutthroat trout; however, additional streams will be sampled in order to 

develop a robust dataset from which changes in stream temperature may be modeled. 

Botanical aquatic habitats (fens, wetlands, riparian areas) can also be directly impacted by predicted 

changes in temperature and precipitation, in much the same way water use values were affected. Predicted 

increases in temperature, associated increases in evapotranspiration, and decreases in aridity indices will 

all result in reducing water availability. Prolonged drought will further reduce groundwater recharge. 

Aquatic habitats in areas where these changes are more pronounced will be most vulnerable. Aquatic 

habitats are currently limited in the drier geographic areas (Uncompahgre, West Elk, Cochetopa) and are 

likely to become even more so. Aquatic habitats on the Grand Mesa may be most vulnerable because the 

aridity index is predicted to drop from above 1 to below 1. 

In reviewing the six previous figures, some areas have high risk much more often than others. Figure 32 

displays a count of how often a given subwatershed has a high risk ranking for the combination of values, 

sensitivities, and stressors. The San Juans geographic area has the largest area (339,717 acres) and largest 

number of subwatersheds (9) that received “High” rankings for all combinations of values, sensitivities, 

and stressors. The Upper Taylor geographic area has the largest area (476,936 acres) of subwatersheds 

with three or more “High” risk rankings. 




Figure 32. Count of High Risk Rankings for Values, Sensitivities and Stressors combined 

Overall vulnerability for the GMUG results from relating the Value × Sensitivity × Stressor Risk rankings 

shown in Figure 32 with the exposure rankings shown in Figure 20. Table 5 combines these rankings. 

Geographic Area 

Exposure 

Ranking* 

Value Risk 

Ranking 

(weighted 

average)** 


Vulnerability 

Ranking** 

Adjusted 

Vulnerability 

Ranking*** 

Uncompahgre 6 1 7/12=0.58 3 

Grand Mesa 5 2 7/12=0.58 4 

San Juans 4 6 10/12=0.83 6 

West Elk 3 3 6/12=0.50 2 

Upper Taylor 2 5 7/12=0.58 5 

Cochetopa 1 4 5/12=0.41 1 

Table 5. Vulnerability Ranking by Geographic Area 

*Exposure Ranking as shown in Figure 20 and Table 4. A ranking of 6 is the highest ranking, and 1 is 

the lowest. Based on greatest change in annual average maximum temperature, annual average 

minimum temperature, and percent change in annual aridity index. 

**Value Risk Ranking as shown in Figure 32. A ranking of 6 is the highest risk to values based on 

weighted average of acres × count of high rankings for each subwatershed. 

***Vulnerability Ranking based (Exposure Ranking + Value Risk Ranking)/12. 

****Adjusted Vulnerability Ranking; Upper Taylor adjusted to be greater than Grand Mesa and 

Uncompahgre because of area in high risk, then Grand Mesa adjusted to be greater than Uncompahgre 

because of higher concentration of values.



APPLICATION 

Data gaps identified in this WVA indicate future inventory needs. More exact locations of road and trail 

crossings can be inventoried. Culverts can be inventoried to determine if they are properly sized for 

potential flood events. Bridges or other crossing structures can be evaluated to determine if they will 

allow debris/sediment/water flow to pass. Crossing inventories should be prioritized in subwatersheds 

with infrastructure at the highest risk and vulnerability. 

Data gaps identified in this WVA indicate future monitoring needs. Stream temperature monitoring can 

be established in those streams of most concern for cutthroat trout, in subwatersheds with the highest risk 

and vulnerability. If strong correlations between increases in air temperature and increases in stream 

temperature can be made, this should identify streams/subwatersheds where cutthroat trout populations 

may be supported in the future 

The WVA can be used to identify where monitoring climate changes (temperature, precipitation, runoff, 

extreme storm events, etc.) can be continued at established weather stations, and expanded into areas 

where climate information is currently extrapolated, to see if predicted changes occur. 

Results from the WVA could be used to identify where predicted changes in runoff overlap with areas 

that have extensive water development, diversion, and allocation. There may be increased pressure to 

enlarge existing developments or construct new storage capacity to capture enough water to meet 

increasing demands downstream in these high use locations. 

The WVA results could be incorporated into future project design and evaluation in those subwatersheds 

that are most vulnerable. Examples include the following. 

• Infrastructure construction/reconstruction in subwatersheds with high risk (sensitivities × 

stressors) may need to be designed to handle higher flood levels or located in less-vulnerable 

areas. 

• Roads should be disconnected from drainage networks. Roads and other manmade features that 

constrain or disconnect channels and floodplains should be removed. 

• Riparian and wetland ecosystems currently in poor ecological health or degraded by loss of 

groundwater should be restored in those subwatersheds/geographic areas expected to become 

more arid. 

• Protect and restore critical or unique habitats that support species survival during critical periods 

(drought, late summer low flows, etc.). 

• The climate change information collected for this WVA can be used in further vulnerability 

assessments of terrestrial resources. 

The WVA can be counted as an accomplishment on the new Performance Scorecard for Implementing the 

Forest Service Climate Change Strategy. 

CRITIQUE 





This watershed vulnerability assessment was focused on water-related resources and did not incorporate 

predicted changes to terrestrial resources, particularly vegetation, and the implications of warmer 

temperatures and potentially reduced precipitation to changes in disturbance regimes (fire, insect, and 

disease), shifts in species composition (increase of invasive species) and the resulting viability of existing 

vegetation communities. Compounding effects on terrestrial ecosystems can have significant influences 

over hydrologic regimes. Similarly, changes in vegetation due to inherent sensitivities (high fire risk) may 

have more influence over watershed conditions than climate changes. 


Climate change reports for the State of Colorado (Ray et al. 2008; Colorado Water Conservation Board 

Draft 2010) provided general statewide projections that also provided information relative to the GMUG. 

Downscaled information and development of two climate change scenarios (Barsugli and Mearns Draft 

2010) served to further describe the range of climate changes that are likely to happen specifically in the 

Gunnison Basin; however, the area where the two scenarios may apply included the entire GMUG Forest 

area. 

VIC data available from the Climate Impacts Group further refined the potential climate changes that may 

occur under several different models. Raster data available at the 6 km-grid scale (approximately) were 

reviewed to see the elevation differences in parameter outputs. Data were also summarized at the HUC-5 

watershed scale. We further summarized data at the geographic-area scale on the GMUG (see Figure 4) to 

see how predicted climate changes might occur on different areas of the Forest that had similar climatic 

regimes. 


Much of the data assembled concerning values, sensitivities, and stressors were limited to that available 

for NFS lands. Some of these data were not complete inventories for the entire GMUG, or the data did not 

portray exact locations (e.g., culvert/crossing locations, stream locations, water rights locations). As a 

result, the composite rankings are more accurate for those subwatersheds (HUC-6) that occur mostly on 

NFS lands, while subwatersheds with larger amounts of off-Forest areas may have erroneous results, 

causing the assessment to compound uncertainties. Collaborative efforts with other agencies and 

landowners/land managers of non-NFS lands within subwatersheds on the GMUG needs to occur so that 

these data gaps can be filled and management implications of climate change can be addressed at a 

complete subwatershed/watershed scale. 

In an effort to save time and build on previous analyses on the GMUG, we used data compiled in 2005 for 

unrelated analyses, and these data were collected at slightly different scales. In some cases, these data are 

no longer current. In others, conversion from the scale used in earlier analyses to the modified 

subwatershed scale used for the WVA was done mathematically, using weighted averages, rather than 

based on spatial data. This introduced further uncertainties into the WVA. 

Because of the inherent sensitivities for erosion/sediment production and runoff response of many 

subwatersheds on the GMUG, the potential effect of extreme storm events is considered to be a big 

vulnerability. There is limited information on extreme storm event frequency and location, and current 

climate change models do not provide projections of storm. 

Baseline stream temperature data is extremely limited, making it hard to interpret what the potential 

effects will be of increased air temperature on stream temperature and changes in cold-water fisheries 

habitat. 





Examples from other units of methods used to deal with different aspects of the analysis were helpful. 

Similarly, examples of vulnerability assessments in general were useful because they provided methods to 

rank different data. 

ArcGIS was the most useful tool to display and evaluate all the spatial data. Microsoft Excel was a useful 

tool to manipulate and summarize tabular data, as well as display modeled outputs. People with expertise 

in these programs are necessary in the team makeup. 


On the GMUG, while we had a relative wealth of information related both to the spatial resource data and 

climate change predictions, we lacked the knowledge to identify and evaluate the implications of 

predicted climate changes to our resource values of concern beyond a very general level. Forests 

completing watershed vulnerability analyses should be teamed up with research station personnel who 

can provide expertise in interpreting the climate change implications portion of the vulnerability 

assessment. It was clear that previous work between the Sawtooth NF and Boise research station had 

created a high level of understanding about the implications of climate change predictions, and familiarity 

with tools available to evaluate where changes are likely to occur and what the impacts of those changes 

may be. 

PROJECT TEAM 

Carol Howe, Resource Information Specialist (GIS), Climate Change Coordinator 

John Almy, Forest Hydrologist 

Clay Speas, Wildlife, Fish and Rare Plants Program Lead 

Warren Young, Forest Soils Scientist 

Ben Stratton, Hydrologist 

Steven Jay, Hydrology Technician 

Sherry Hazelhurst, Deputy Forest Supervisor 


Carol Howe 

Grand Mesa, Uncompahgre and Gunnison National Forests 

2250 Hwy 50 

Delta CO, 81416 

970-874-6647 

chowe@fs.fed.us 

REFERENCES 

Barsugli, J.J. and L.O. Mearns. Draft 2010. Climate and Hydrologic Change Scenarios for the Upper 

Gunnison River, Colorado. Prepared for The Nature Conservancy in support of the southwest Climate 

Change Initiative’s Climate Change Adaptation Workshop for Natural Resource Managers in the 

Gunnison Basin. 




Christensen, N.S., and D.P. Lettenmaier. 2007. “A multimodel ensemble approach to assessment of 

climate change impacts on the hydrology and water resources of the Colorado River Basin.” Hydrol. 

Earth Syst. Sci., 11, 1417–1434 (www.hydrol-‐earth-‐syst-‐sci.net/11/1417/2007/).” 

Colorado Water Conservation Board. Draft 2010. Colorado River Water Availability Study; Phase I 

Report. 

Furniss, M.J., B.P. Staab, S. Hazelhurst, C.F. Clifton, K.B. Roby, B.L. Ilhadrt, E.B. Larry, A.H. 

Todd, L.M. Reid, S.J. Hines, K.A. Bennett , C.H. Luce, P.J. Edwards. 2010. Water, climate change 

and forests: watershed stewardship for a changing climate. Gen. Tech. Rep. PNW_GTR-812. USDA 

Forest Service, Pacific Northwest Research Station. Portland OR. 

Hirsch, C.L., S. E. Albeke, and T. P. Nessler. 2006. Range-wide status of Colorado River cutthroat 

trout (Oncorhynchus clarkii pleuriticus): 2005. Colorado Division of Wildlife, Denver, CO. 

IPCC. 2008. Technical Paper of the Intergovernmental Panel on Climate Change on Climate Change and 

Water. [Bates, B.C., Z.W. Kundzewicz, S.Wu, and J.P. Palutikof, (eds.)] IPCC Secretariate, Geneva. 

(Available at: http://ipcc.ch/pdf/technical-‐papers/climate-‐change-‐water-‐en.pdf). 

Painter, T.H., J. Deems, J. Belnap, A. Hamlet, C.C. Landry, and B. Udall. 2010. Response of 

Colorado river runoff to dust radiative forcing in snow. Proceedings of the North Academy of Sciences. 

(accessed at www.pnas.org/content/early/2010/09/14/0913139107.full.pdf+html) 

Ray, A.J., J.J. Barsugli, K.B. Averyt, K. Wolter, M. Hoerling, N. Doesken, B. Udall, R.S. Webb. 

2008. Climate Change in Colorado: a Synthesis to Support Water Resources Management and 

Adaptation. Western Water Assessment. Boulder, CO. 

Rieman, B.E. and D.J. Isaak. 2010. Climate Change, Aquatic Ecosystems, and Fishes in the Rocky 

Mountain West: implications and alternatives for management. Gen. Tech. Rep. RMRS-GTR-250. USDA 

Forest Service, Rocky Mountain Research Station. Fort Collins, CO. 

Spears, M., L. Brekke, A. Harrison, and J Lyons. 2009. Literature Synthesis on Climate Change 

Implications for Reclamation’s Water Resources. Technical memorandum 86-68210-091. U.S. 

Department of the Interior, Bureau of Reclamation, Research and Development Office. Denver, CO. 




White River National Forest 

March, 2012 

Prepared by: 

Mark Weinhold 

Forest Hydrologist 

White River National Forest 

Glenwood Springs, Colorado 


White River National Forest Watershed Vulnerability Assessment, Rocky Mountain Region (R2) 

BACKGROUND 

The White River National Forest is located in west central Colorado, on the western slope of the Rocky 

Mountains in the Rocky Mountain Region (R2) of the USFS. Over the 2.3 million acre forest, elevations 

start from a low of about 5,500 feet and rise to include several peaks over 14,000 feet. Glaciation has 

shaped the higher elevations. Granitic rocks are prevalent on the eastern side of the forest; sedimentary 

formations dominate the western side. Most of the precipitation falls as snow in the winter, although 

summer thunderstorms are common. Snowmelt from the forest into the Colorado River provides water to 

27 million people in 7 states and two countries (Painter et al. 2010). Peak flows are generally associated 

with snowmelt, except for the western edge of the forest. 

The White River is the most visited National Forest in the country, largely because of winter sports. Most 

of Colorado’s largest ski areas (Vail, Keystone, Breckenridge, Aspen, etc.) are permit holders on the 

Forest. Consequently, there is a keen interest in how a changing climate may affect air temperatures and 


INTRODUCTION 

Aquatic biological systems, such as those supported by National Forests, have evolved under certain 

climatic conditions. As the climate changes, it is reasonable to anticipate that a watershed’s ecological or 

biological values could also change. The analysis described herein is an attempt to apply expected 

changes in climate to large portions of the landscape, and determine which areas (and their associated 

resource values) are least resilient and therefore most susceptible to adverse effects from a changing 

climate. 

The objective of this effort is to define a process that sorts blocks of the landscape (HUC-6 subwatersheds 

in this case) into categories that express their relative vulnerability to climate change. By way of analogy, 

we propose to take all the subwatersheds on the forest and (mentally) shake them through a series of 

sieves in order to identify those that have the least resiliency to the anticipated changes in temperature, 

precipitation, and runoff. 

Because this process is intended to cover large landscapes (2.3 million acres in this case), it is necessary 

to rely on existing data. The GIS queries that make up the basis for the assessment rely on common 

corporate layers from either the Forest Service or state agencies. 

A key step at the outset of this process was the identification of an appropriate scale of analysis. Since 

the analysis is aquatics-based, watershed boundaries were chosen. Because subwatersheds generally 

coincide with the management scale of most Forest activities, and are also small enough to allow local 

expression of factors such as aspect, elevation, vegetation type, etc., they were chosen as the unit of 

analysis. 

The schematic in Figure 1 shows the general thought process behind the analysis protocol. Resource 

values (for example, a sensitive species of trout), are supported by a complex interaction of ecological 

landscape-scale drivers. These drivers define the ecological context (environment) of the watershed and 

can include such attributes as geology, aspect, precipitation, and glaciation, etc. Changes to this 

environment occur constantly, but large changes from anthropogenic or climatic stressors may affect the 

resiliency of the resource value of concern. Determining how these ecological and anthropogenic 

characteristics interact with anticipated climatic stressors to affect the relative resiliency of each 

subwatershed is the objective of this analysis. 



Figure 1. Schematic of the climate change vulnerability assessment process 

ANALYSIS PROCESS 

Determination of the relative vulnerability of each subwatershed involves the following steps, which are 

discussed in detail below: 1) identify the aquatic resource values of concern; 2) quantify the anticipated 

exposure from a changing climate; 3) identify the relative influence of the ecological drivers and 

anthropogenic influences for each subwatershed; and 4) assess the relative vulnerability of the resource 

values based on the interaction of the ecological drivers, anthropogenic influences, and the anticipated 

climate change exposure. 

Step 1. Identify the Resource Values of Concern 

Initial brainstorming on prominent aquatic resources gave a laundry list of potential values. These 

included aquatic habitat, water uses, infrastructure (roads, trails, and campgrounds) in streams or 

floodplains, wetlands, and water dependent recreation. This list proved to be overly ambitious and was 

eventually pared down. The final list of aquatic resource values to be considered includes the following. 

1. Aquatic Habitat - specifically for Colorado River cutthroat trout and boreal toads 

2. Water Uses - irrigation and water supply 

3. Infrastructure - culverts and bridges at road-stream crossings 



This abbreviated list was considered comprehensive enough to cover the most significant aquatic issues 

while not generating redundant information across a long list of resource values. It became apparent that 

narrowing the list of resource values was justified since there is only modest variability in the final 

relative vulnerability of the three selected resource values. 

Step 2. Quantify the Anticipated Exposure from Climate Change 

Exposure is the term used to describe the amount of anticipated change in climate over time. The types of 

exposure typically considered for the mountainous West include changes in air temperature, changes in 

precipitation, and changes in runoff. 

Exposure estimates are not only highly variable but are highly uncertain as well. Variability of exposure 

estimates arise primarily from differences in carbon emission scenarios and the time frame of concern. 

High (A2) and low (B1) emission scenarios give very different exposure results when modeled at midcentury 

(often shown as year 2040 or 2050) versus those modeled at the end of the century. 

Uncertainty is also a major factor in estimating exposure. Exposure estimates, whether for temperature, 

precipitation, or runoff, are generated from global circulation models that attempt to predict weather 

patterns around the globe simultaneously for any given emission scenario. These large-scale global 

estimates are then down-scaled to smaller areas of concern, such as a state or some smaller region. A 

single model is rarely used to estimate exposure in a given locale. Rather, many different models are run 

and the exposure value presented is often the median of the predictions, along with a potential range of 

values. 

Since water supply is such a significant issue in the arid west, many states have compiled summaries of 

climate change predictions in order to assess future water supplies. Colorado is one of those states. For 

this analysis, climate change exposure data were taken from the 2008 report for the Colorado Water 

Conservation Boards entitled Climate Change in Colorado: A Synthesis to Support Water Resources 

Management and Adaptation (Ray et al. 2008). 

Predicted changes to temperature, precipitation, snowpack, and runoff (Christensen and Lettenmaire, 

2006) are shown below in Figures 2 through 5. Figure 2 shows that air temperatures are predicted to 

increase over time. For the high-emission scenario (A2), the median predictions suggest an increase of 2.5 

to 4.5 degrees Fahrenheit for mid- and late-century timeframes, respectively. This is in addition to an 

estimated 2 degree increase that has occurred over the last 30 years. Summers are projected to warm more 

than winters; winter projections show fewer extreme cold months, more extreme warm months, and more 

strings of consecutive warm winters (Ray et al. 2008). These warmer temperatures are likely to influence 

precipitation type, stream temperatures, and stream flow rates. 



Figure 2. Possible air temperature changes predicted from down-scaled global circulation models for the Colorado 

River basin (Christensen and Lettenmaier 2006) 

Figure 3 shows predicted changes in precipitation relative to the long-term historical record. Given the 

variability of the predictions, no consistent trend in annual precipitation is evident. However, other 

research has shown that shifts in the type of precipitation (primarily snow to rain) and shifts in the 

seasonal distribution are likely (Ray et al. 2008). 



Figure 3. Possible changes in annual precipitation predicted from down-scaled global circulation models for the 

Colorado River basin (Christensen and Lettenmaier 2006) 

Regarding precipitation, of particular interest is the change in snowpack with elevation. Figure 4 shows 

results from Christensen and Lettenmaier (2006), which suggest that snowpacks are expected to decline at 

elevations below about 8,500 feet. In western Colorado, the current transition from a rain-snow 

dominated precipitation regime to a snow-dominated regime occurs at around 7,500 feet elevation. This 

transition elevation is expected to rise with time and emissions. For this analysis, we considered the 

elevation band from 7,500 to 8,500 feet elevation to include snowpack at risk. That is, we expect more of 

the precipitation to occur as rainfall, as opposed to snow, which would affect both the timing and 

magnitude of streamflow. 



Figure 4. Predicted changes in Colorado River basin snowpack (Christensen and Lettenmaier 2006) 

Lastly, Figure 5 shows the predicted decrease in annual runoff for the Colorado River Basin. Median 

estimates from the multi-model runs approach 10% by mid and late century. Multiple studies in the 

Colorado River basin show predicted decreases in runoff between 6% and 20% by 2050 (Ray et al. 2008). 

Lower runoff is also coupled with a shift in the peak flow hydrograph. The peak is anticipated to occur 

earlier by two to four weeks, perhaps more, depending on the influence of dust on the snow surface. 

Recent research in Colorado has suggested that peak flows occur up to 3 weeks earlier than they did 

historically. This is at least partially due to dust layers on the snow surface that reduce snow reflectivity 

and increase the amount of solar radiation absorbed in the snowpack (Painter et al. 2010). Thus, not only 

will there be less water in streams and available for water uses, but the peaks flows will likely be 

occurring before the irrigation season begins. This would surely lead to an increase in the number of 

proposals for water storage. 



Figure 5. Possible runoff changes predicted from down-scaled global circulation models for the Colorado River 

basin (Christensen and Lettenmaier 2006) 

In summary, there are three potential outcomes of the anticipated climate change exposure. Most 

importantly, runoff volumes are likely to decrease, potentially exacerbating low flow conditions. This 

would likely be accompanied by higher water demand for irrigation, associated with higher air 

temperatures. All signs suggest an inevitable conflict between the Aquatic Habitat and Water Uses 

resource values. 

Secondly, although the published exposure data make little reference to flood events, there appears to be a 

trend toward more extreme weather events. The possibility of higher and more frequent flood events 

would have a direct impact on the Infrastructure/roads resource value. 

Lastly, as noted previously, we have seen average air temperatures increase over the last 30 years, and the 

data suggest a continuation of that trend. This would logically translate to increases in stream 

temperatures. However, since Colorado River cutthroat are typically pushed to the upper limits of their 

range through competition with brook trout, their reproductive success can be limited by cold water 

temperatures. In this rare case, an increase in stream temperatures may actually work to their benefit. For 



this reason, projected increases in stream temperatures are not carried forward in this process as a 

potential impact. 

Step 3. Identify Landscape-Scale Ecological and Anthropogenic Drivers 

At this point in the analysis, we have a general idea about the magnitude and direction of effects to 

aquatic systems from climate change. From the exposure data, we can see that temperatures will increase, 

some elevations will experience more rain than snow, and runoff timing may shift earlier while overall 

volume may decrease. With these potential changes in mind, we looked at the landscape-level drivers, 

both inherent to the subwatershed and human-created, that could either exacerbate or buffer these effects. 

Inherent Attributes of the Project Area Subwatersheds 

The resiliency of a watershed to any change is largely a function of parent geology, typical climate, 

topography, and vegetation. For this analysis, these factors were subdivided into more specific attributes 

that could be queried in GIS by subwatershed. The attributes considered most important for the White 

River National Forest are as follows: 

Geochemistry of the parent geology. Aquatic systems are intimately linked with the chemistry of the 

parent geology. In particular, calcareous geologies contain calcium carbonate (CaCO3), which dissolves 

to form ions that influence primary productivity in a stream. The weathering of these rocks also raises the 

stream pH and produces carbon dioxide for photosynthesis (Staley 2008). Because of the buffering effects 

to aquatic ecosystems from increased productivity, the percentage of a subwatershed with calcareous 

parent geology was used as a measure of resiliency to climate change. 

Extent of glaciation. Glacial processes have made some landscapes more suitable for wetland and 

riparian area developments by flattening the gradient of high mountain valleys and slowing runoff. 

Lateral and terminal moraines have created topography that encourages the slow movement and retention 

of large volumes of snowmelt-recharged groundwater. Consequently, glaciated environments typically 

have the highest densities of high-quality wetlands on the forest. Since glaciation generally led to a 

significant local influence on water availability and distribution, the percent of a subwatershed that was 

glaciated is used as a measure of inherent resiliency to climate change. 

Aspect. In snow dominated systems, aspect is a key factor affecting the size and longevity of the 

snowpack. South aspects tend to lose snow to evaporation or sublimation, even in the middle of winter. 

Subwatersheds dominated by southern aspects are expected to carry less snow for shorter periods under a 

warming climate scenario. Therefore, the percent of a subwatershed with a south, southeast, or southwest 

aspect is used as a measure of inherent resiliency to climate change. 

Hydroclimatic regime. This refers to the typical precipitation regime for a subwatershed. In the central 

Colorado Rocky Mountains, landscapes below about 7,500 feet typically have much of their precipitation 

and storm peaks associated with rainfall. Landscapes above about 7,500 feet in elevation typically have 

most of their precipitation and storm peaks associated with snowfall and snowmelt. As the climate warms, 

we expect that the transition from a snow-dominated to rain-snow-dominated precipitation regime will 

migrate upslope. The elevation band from 7,500 to 8,500 feet is considered to be an at-risk zone for 

snowpack. For this analysis, the percent of a subwatershed within the at-risk snow elevation band is used 

as a measure of inherent resiliency. 

Weighted precipitation. This attribute refers to the amount of precipitation that falls on the landscape as 

either snow or rain. In the Rocky Mountains, precipitation amount varies significantly with elevation and 

orographic effects. The Parameter-elevation Regressions on Independent Slopes Model (PRISM) database 



available from Oregon State University was used to determine composite precipitation values for each 

subwatershed, weighted by elevation. Since the amount of precipitation a subwatershed receives has a 

direct effect on aquatic ecosystems, weighted precipitation is used as a measure on inherent resiliency. 

Extent of surface water features. Groundwater movement and storage plays a large role in maintaining 

streamflow and stream temperatures. We found that the parent geology was not necessarily a reasonable 

predictor of shallow groundwater that regularly interacts with surface water. Instead, the presence of 

surface water and springs from the National Hydrography Dataset (NHD) GIS layers was used to estimate 

the percentage of a subwatershed with surface water or springs. Because of the buffering effects shallow 

groundwater has on aquatic ecosystems, this attribute was also used as a measure of inherent resiliency. 

Extent of large-scale pine beetle mortality. In snow-dominated systems, vegetation locally affects 

hydrology through evapotranspiration, canopy interception, and extent of snow scour. As the pine beetle 

epidemic progresses across western Colorado, we expect to see less evapotranspiration, less canopy 

interception, and more redistribution of snow as forest openings increase. Because of these effects on the 

annual hydrograph, the percentage of a subwatershed affected by pine beetle mortality was used as a 

measure of resiliency to a changing climate. 

Anthropogenic Influences in the Project Area Subwatersheds 

Human influences can also affect the resiliency of a subwatershed, depending on the amount of 

management-related activity that occurs. For the White River National Forest, the following 

anthropogenic influences were considered to have potentially significant effects on aquatic resources: 

Water uses. The amount of water withdrawn from a steam has a direct effect on the health of the aquatic 

system. The more water that is withdrawn, the more stress a system is exposed to and the less resilient it 

is to additional changes in water supply. Additionally, changes in streamflow have been associated with a 

competitive advantage for invasive species (Merritt and Poff 2010). In order to capture the cumulative 

change to the natural hydrology, the number of diversions per square mile was used as a measure of 

resiliency to climate change. 

Development (primarily roads). Roads and road ditches can have significant effects on how water is 

routed across the landscape. Ditches collect surface water (or intercept shallow subsurface water) on hill 

slopes, and act as tributary extensions of the stream network. Routing water off the landscape more 

quickly would have the net effect of exacerbating anticipated effects of climate change on runoff. In order 

to capture the influence of roads on the stream network, the road density, calculated as miles per square 

mile, was used as a measure of resiliency to climate change. 

Extent of beetle salvage. Performing salvage logging operations to remove standing dead trees can have 

additional effects on watershed hydrology. First, removing standing dead trees further reduces the 

interception of snow and can increase snow scour as openings increase in size. Additionally, most logging 

operations typically involve some new roads, at least temporarily. These effects may be slightly buffered 

in the long term since removal of trees may allow for quicker reforestation and subsequent hydrologic 

recovery. The percentage of a watershed proposed for salvage logging was used as a measure of 

resiliency to climate change. 

Step 4. Assess the Relative Vulnerability of the Resource Values 

In order for the relative vulnerability among subwatersheds to be determined, each inherent and 

anthropogenic attribute needs to be broken into categories of high, medium or low. Then each attribute 



needs to be weighted in order to combine them into a meaningful aggregate score. The processes for 

assigning categories and relative weights are as follows. 

Determination of High, Moderate, and Low Categories for Subwatershed Attributes 

In order to apply a simple mathematical ranking system by subwatershed, each of the previously 

discussed attributes required binning into categories. The amount of influence that an attribute exerts 

within a given subwatershed was categorized as high, moderate, or low. 

Upon inspection, most of the attributes or influences have no physical threshold to suggest a breakpoint 

between categories. For example, we don’t have any data to suggest how many diversions per square mile 

a subwatershed can contain and still have a low influence on aquatic systems. Since the objective of this 

analysis was to determine relative vulnerability between subwatersheds, a simple and objective approach 

was used. For each attribute listed, the distribution of the 166 subwatersheds was plotted and the quartiles 

determined. By definition, the first quartile is the 25 th percentile of the ranked data, the second quartile is 

the median, and the third quartile is the 75 th percentile of the ranked data. Subwatersheds below the first 

quartile (lowest 25%) were ranked as low influence; subwatersheds between the first and third quartile 

(middle 50%) were ranked as moderate; subwatersheds above the third quartile (top 25%) were ranked as 

high. See the example plot for road density below in Figure 6. 

Figure 6. A sample histogram of diversions per square mile across all subwatersheds, and the use of quartiles to 

categorize the relative influence on resiliency as high, moderate, or low 

Determination of the Relative Weights of Inherent and Anthropogenic Attributes 

While each of the attributes listed has some effect on the ultimate resiliency of the subwatersheds, they do 

not have equal effects. For example, the amount of precipitation or the amount of water withdrawn from a 

subwatershed is likely more important than a primary productivity increase from calcareous geology. 

Consequently, a simple method of scaling the relative influence of the attributes was developed. 

Since physically removing water from the stream (water uses) has the most direct effect on aquatic 

systems, each attribute was weighted relative to that, with values ranging from 0.25 (1/4 the effect) to 1 

(similar effect). The assigned weights are as follows: geochemistry of parent geology (0.25), extent of 



glaciation (0.75), south aspect (0.50), hydroclimatic regime (1.0), weighted precipitation (1.0), extent of 

surface water features (1.0), extent of pine beetle mortality (0.5), water uses (1.0), development/roads 

(0.5), and the extent of beetle salvage (0.5). 

Determination of a Summary Numeric Ranking for Each Subwatershed 

At this point, the seven natural and three anthropogenic attributes that could either add to or buffer the 

expected climate change effects have been identified. These factors have also been categorized as having 

a high, moderate, or low influence and they have been weighted based on the relative strength of their 

influence. See the summary in Table 1 below. 

In order to aggregate these factors into a single rating, a simple numeric scheme was used. Factors 

exerting a high influence were assigned a value of 5, medium a value of 3, and low a value of 1. The 

score for each attribute was multiplied by the weighting factor, and those products were averaged for all 

attributes within a subwatershed. 

Once the average score was calculated for all the subwatersheds, they could easily be partitioned into 

groups based on their numeric ‘vulnerability.’ Again, given that no actual physical/biological thresholds 

exist based on the numbering scheme used, quartiles served as a consistent and systematic way to 

categorize subwatersheds with high, moderate, and low risk of impacts from climate change. Of all the 

166 subwatersheds evaluated, the 25% with the highest overall scores were ranked as high vulnerability. 

The 25% with the lowest overall scores were ranked as low vulnerability. The middle 50% were ranked as 

moderate vulnerability. 

Subwatershed Attribute Name Type of Attribute 

Relative 

Weight 

Net Effect Relative to 

Climate Change 

Geochemistry of parent geology Inherent to watershed 0.25 Buffer 

Extent of glaciation Inherent to watershed 0.75 Buffer 

Aspect Inherent to watershed 0.50 Additive 

Hydroclimatic regime Inherent to watershed 1.0 Additive 

Weighted precipitation Inherent to watershed 1.0 Buffer 

Extent of surface water features Inherent to watershed 1.0 Buffer 

Extent of large-scale pine beetle mortality Inherent to watershed 0.5 Buffer (short term) 

Water uses Anthropogenic 1.0 Additive 

Development (primarily roads) Anthropogenic 0.5 Additive 

Extent of beetle salvage Anthropogenic 0.5 Additive (short term) 

Table 1. Summary of attribute types affecting subwatershed resiliency to climate change 

Presentation of Results 

Recall that the subwatershed attributes were rated based on their effect on one of the resource values 

(Aquatic Habitat, Water Uses, or Infrastructure/roads). Consequently, the previously described steps had 

to be repeated for each resource value. The results are graphically shown in Figures 7 through 9. 

Note that the presence or absence of the resource value did not play a role in the numeric ranking and 

categorization. Rather, the subwatershed’s vulnerability was assessed based on the natural and 

anthropogenic attributes, then the known resource value occurrences were overlaid on top of those 

ratings. In this case, the mapped elements included Colorado River cutthroat trout and boreal toad 

populations for the Aquatic Habitat resource value, points of diversion for Water Uses resource value, and 

road-stream crossing locations for the Infrastructure/Roads resource value. Therefore, areas of initial 

concern for managers would be those subwatersheds with high vulnerability AND a high concentration of 

the resource value. 



Figure 7. Climate change vulnerability rating for the Aquatic Habitat resource value. Red shading depicts 

subwatersheds with the highest vulnerability. Cutthroat trout and boreal toad populations are shown as green lines 

and green dots, respectively. 



Figure 8. Climate change vulnerability rating for the Water Uses resource value. Red shading depicts 

subwatersheds with the highest vulnerability. Points of diversion for water uses are shown as black dots. 



Figure 9. Climate change vulnerability rating for the Infrastructure/Roads resource value. Red shading depicts 

subwatersheds with the highest vulnerability. Road-stream crossings are shown as blue dots. 

As expected, the lower elevation subwatersheds are those that display the highest vulnerability to a 

changing climate. These are the watersheds with lower precipitation, more area in the rain-snow transition 

zone, and an absence of glaciated terrain. Because of the low elevation, these subwatersheds also tend to 

have a large private-land component and the highest number of irrigation diversions. 

Note that even as the resource value changes, there is not a huge variability in the mapped outcome. The 

natural and anthropogenic factors do not radically change, which supports the notion of minimizing the 

number of resource values considered. In this case, two resource values areas could have sufficed: One 

that captures effects from decreasing low flows (droughts) and one that captures increasing high flows 

(floods). 

APPLICATION 

Focus on Anthropogenic Influences 

As a whole, management activities on National Forests don’t create a lot of greenhouse gasses. So instead 

of focusing on the causes of climate change, our concern might center on increasing the resiliency of our 

landscapes to minimize their negative response to climate change. Looking back at the analysis process 

used, our role in increasing resiliency is ultimately very narrow, because much of a subwatershed’s 



sensitivity is an artifact of its inherent characteristics, such as geology, elevation, precipitation, etc. In 

other words, we can’t affect most of the attributes that influence resiliency. Therefore, the focus narrows 

to the few things that management can actually affect – the anthropogenic influences such as water uses, 

roads, and vegetation management. 

In the subwatersheds with the highest sensitivities, any activity that maintains or increases water quantity 

or runoff timing would ultimately be beneficial. Specific actions could include contesting new water 

rights proposals, exploring ways to convert existing water rights into instream flows, and anticipating 

storage proposals (which are likely to increase in both size and frequency). 

This analysis could also help guide implementation of our travel management plan by directing where 

roads should be decommissioned or where reconstruction/maintenance should be scheduled to 

hydrologically disconnect roads from the stream network. Similarly, this analysis could also help 

prioritize aquatic organism passage projects at road-stream crossings to ensure that aquatic residents are 

able to migrate to suitable habitat as streamflow and temperatures change. Selecting the subset of high 

vulnerability watersheds in high pine beetle mortality areas would also help prioritize road-stream 

crossings for upgrades relative to floods and debris. 

Lastly, with a half million acres of pine beetle mortality on the Forest, the results of this analysis could 

help direct where active vegetation management could benefit the recovery process by enhancing natural 

reproduction, hydrologic recovery, stream shading, and future large woody debris recruitment. 

Integration with the Watershed Condition Framework Process 

The recently completed process for the watershed condition assessment ended with a condition rating for 

each subwatershed on the forest. There were 12 attributes that were rated, but the following subset of 

those could be directly affected by climate change: 

• 1.2 - Water Quality Problems 

• 2.1 - Water Quantity 

• 4 - Aquatic Biota (Exotics and Invasives) 

• 10.1 - Vegetation Condition 

• 12 - Forest Health (Insects and Disease) 

Changes in runoff from climate change would have direct effects on water quantity (attribute 2.1), and 

indirect effects on water quality (attribute 1.2) as dilution flows diminish. Less runoff may also mean 

more indirect effects on aquatic and riparian biota (attribute 4.0), because exotic species tend to compete 

well in environments with modified flows and temperatures. 

Changes in air temperature and the distribution of precipitation types would eventually affect the 

distribution of vegetation types and the overall vegetation condition (attribute 10.1). Local experience 

with the mountain pine beetle has shown that insects and diseases (attribute 12) can propagate in 

unexpected ways with small changes in air temperature. 

Since the Watershed Condition Framework assessment and this climate change vulnerability assessment 

were both conducted at the subwatershed scale, they are easily integrated. Identifying areas where 

diminished watershed condition attributes overlap with high climate change vulnerability can help target 

restoration priorities. 



LESSONS LEARNED 

The important thing to remember is that this analysis is an attempt to determine the relative vulnerability 

of subwatersheds to the anticipated effects of climate change and to give managers a general idea about 

geographic areas of concern. It is, by nature, a broad-brush approach, and the level of precision and detail 

of the input parameters need to be commensurate with the precision of the final product. To a significant 

degree, less is more. 

As an example, when the scope of the analysis is being determined, there is inevitably a lot of 

brainstorming about what resource values would be affected by certain aspects of a changing climate. The 

initial list of resource values can be long. We found that resource issues often had similar sensitivities and 

expected responses. For example, two resource values that both respond negatively to decreases in 

streamflow are likely to give very similar vulnerability results. In the mountainous region of the Rocky 

Mountain west, it may be reasonable to limit resource values to one affected by timing/magnitude of 

decreasing flows, one affected by timing/magnitude of increasing peak flows, and/or one affected by 

changes in stream temperatures. 

Similarly, the list of inherent subwatershed attributes and anthropogenic influences (e.g., geology, 

precipitation, roads) that affect the vulnerability of a resource value can also be quite long. Although 

many small factors can cumulatively affect resource value vulnerability, they may not exert much 

influence in a particular numeric rating scheme. We found that factors with a low influence (assigned 

weights) had very little influence on the final rating. It would be a simple matter to do a sensitivity 

analysis of the numeric results to see if some attributes could be dropped early in the process, to 

streamline the analyses. 

Finally, as time goes on, much more detailed data on climate change exposure becomes available. Models 

are constantly being tuned and down-scaled to smaller areas. These data have limits based on their 

uncertainty, and that uncertainty grows with down-scaling. We structured this analysis so that the actual 

values for temperature changes, runoff changes, etc. were not critical. Rather, we focused on the 

magnitude and direction of the predicted change. We were more concerned with the direction of change 

(increasing or decreasing) and whether that change was a big number or small number relative to the 

annual variability that we see now. Not getting lost in the myriad of details of the Variable Infiltration 

Capacity model results was easily justified by keeping the original goal of the process in mind. 

REFERENCES 

Christensen, N. and D.P. Lettenmaier. 2006. A multimodel ensemble approach to assessment of 

climate change impacts on the hydrology and water resources of the Colorado River basin. Hydrology and 

Earth System Sciences Discussion, 3:1-44. 

Merritt, D.M. and N.L. Poff. 2010. Shifting dominance of riparian Populus and Tamarix along gradients 

of flow alteration in western North American rivers. Ecological Applications, 20(1): 135-152. 

Painter, T.H., J.S. Deems, J. Belnap, A.F. Hamlet, C.C. Landry and B. Udall. 2010. Response of 

Colorado River runoff to dust radiative forcing in snow. Proceedings of the National Academy of 

Sciences of the United States of America. 6 pp. Full report available at 

http://www.pnas.org/content/early/2010/09/14/0913139107.full.pdf+html 



Ray, R.J, J.J Barsugli, and K.B. Averyt. 2008. Climate change in Colorado: a synthesis to support 

water resources management and adaptation. A report by the Western Water Assessment for the Colorado 

Water Conservation Board. 52 pp. Full report available at 

http://wwa.colorado.edu/CO_Climate_Report/index.html 

Staley, D.M., D.J. Cooper, and E. Wohl, 2008. White River National Forest – Aquatic and Riparian 

and Wetland Assessment. Draft Report, USDA Forest Service, Rocky Mountain Region. Denver, CO. 




Coconino National Forest 

April 2012 

Prepared by: 

Rory Steinke, CPSSc 

Watershed Program Manager 

Coconino National Forest 

Flagstaff, Arizona 


Coconino National Forest Watershed Vulnerability Assessment, Southwest Region (R3) 

INTRODUCTION 

This report presents the results of a Watershed Vulnerability Assessment (WVA) conducted on the 

Coconino National Forest (CNF) during 2010 and 2011. The Forest is located in Arizona in the Southwest 

Region (R3) of the USFS. The CNF volunteered to participate in a collaborative project between USFS 

and FS Research to develop processes to assess watershed climate vulnerability. 

The objective of the assessment was to evaluate the relative vulnerabilities of watersheds to hydrologic 

changes that could result from a changing climate. 

The pilot assessment process employed a very simple model of vulnerability, based on the combination of 

values at risk, the sensitivity of those values to change, and the potential for exposure. The model is 

illustrated in Figure 1. 

Figure 1. Conceptual Model for Assessing Watershed Vulnerability 

The pilot team also established a step-wise approach to the vulnerability assessment. The process is 

patterned after Watershed Analysis (USDA, 1994). The organization of this report follows the WVA 

process steps, which are as follows. 

• Step 1 - Establish the Scope and Water Resource Values that Will Drive the Assessment 

• Step 2 - Assess Exposure 

• Step 3 - Assess Watershed Sensitivity and Watershed Condition 

• Step 4 - Evaluate and Categorize Vulnerability 

• Step 5 - Response and Recommendations for Making WVA Useful for Managers 

• Step 6 - Critique the Vulnerability Assessment 

STEP 1 - Establish the Scope and Water Resource Values that Will Drive the Assessment 

Five fifth-field watersheds on the Forest were selected for analysis. These watersheds were selected 

because they support most of the aquatic resource values on the Forest. The watersheds are listed in Table 

1, and displayed in Figure 2. 



Watershed HUC 

Upper Clear Creek 1502000803 

West Clear Creek 1506020301 

Fossil Creek 1506020303 

Beaver Creek 1506020206 

Oak Creek 1506020205 

Table 1. Watersheds on the Coconino NF included in the Watershed Vulnerability Assessment 

Figure 2. Watersheds included in the Coconino NF Watershed Vulnerability Assessment 

Water is an extremely important resource on the CNF. Parts of the Forest lie within the Central Highlands 

of Arizona. This area receives higher precipitation than most of the state, and therefore is an important 

source of runoff and groundwater, locally and regionally (Figure 3). Water from the watersheds selected 

for the assessment supports a variety of important aquatic resources that include both natural systems and 

human uses. Perennial water is relatively scarce, and demands for both instream uses and diverted water 

are high. 



Figure 3. Annual precipitation for Arizona. Coconino NF and selected assessment watersheds include areas 

of relatively high precipitation for the region (from NOAA, 1994). 

Habitat degradation and competition with invasive species have severely restricted the distribution of 

numerous aquatic species. The regional human population continues to grow, as does demand for water. 

Competing demands for water will continue, and these demands are likely to be exacerbated by climate 

change. The National WVA pilot proposed that aquatic species, water uses, and infrastructure be included 

in each assessment. The CNF assessment included those values as well as two other resource values 

riparian and spring habitats, stream habitat) in the assessment. Each resource is briefly described below. 



Figures 4 a-b. Historic and existing distribution of selected aquatic species on the Coconino NF 

Native Aquatic Species 

The CNF supports a wide variety of native aquatic species. The distribution of these species has been 

greatly reduced due to water development, degraded habitat, and invasive non-native species (see Figures 

4 a-b). Species in the analysis include both native warm water fishes and herpetiles. 

The CNF is home to an extensive list of Threatened, Endangered and Sensitive (TES) fish species. The 

fisheries biologist selected four fish species for inclusion in the analysis, all of which are currently present 

in subwatersheds within the analysis area (rather than downstream). The species selected for inclusion are 

listed in Table 2. Several are listed under the Endangered Species Act, and on the CNF, some are 

currently found only in the analysis area. 

Four other listed, candidate or species of concern were included as resources in initial assessment efforts 

but not carried forward due to their very limited distribution and co-location with other species. These 

were Gila Trout (re-introductions of the species on CNF have been discussed), Red Rock Stone fly, and 

the Fossil Springs and Page Springs spring snails. 



Species Species Status 

Amphibian Species 

Chiricahua Leopard Frog Threatened 

Lowland Leopard Frog Sensitive 

Northern Leopard Frog Sensitive 

Arizona Toad Sensitive 

Reptiles Species 

Narrow-headed Garter Snake Sensitive 

Mexican Garter Snake Sensitive 

Warm Water Fish Species 

Little Colorado Spine Dace Threatened 

Gila Chub Endangered 

Loach Minnow Threatened 

Spikedace Threatened 

Table 2. Aquatic species (and their status) included in the analysis 

For the analysis, resource value was rated based on the number of herpetile species present in each 

watershed. Likewise, the number of the four warm-water fish species found in each subwatershed was 

used to rate the resource value. Results of these ratings are shown in Figures 5 a-b. 



Figures 5 a-b. Location of selected herpetile and warm water fish species 




The Forest has a relatively high density of roads, with associated stream crossings. Several campgrounds 

are located within or adjacent to floodplains and may be susceptible to flood damage. In addition, 

numerous forest service roads, county roads, and state highways are located adjacent to stream channels 

and may be vulnerable to flooding. Characterization of the value of each subwatershed (HUC-6) for the 

resource was based on the density of road crossings (data source: Forest road route and stream route 

layers). Frequency distribution of the sixth field densities was used to rate each watershed as high, 

moderate, or low. This rating was made after analysis of both channel crossings and miles of road within 

150 ft of channels. Results showed a very high correlation (>0.90) between the frequency of road 

crossings and the miles of road within 150 feet of channels. It was assumed that including the miles of 

adjacent roads added little to the analysis, so the road crossing data were used for the infrastructure 

resource ratings. 

Figure 6. Density of road stream crossings and location of campgrounds (red triangles) within 300 ft of stream 

channels. Darker colors represent highest density; grey indicates lack of data. 

Campgrounds located within 300 ft of a channel were also considered (see Figure 6). Campgrounds were 

not included in the infrastructure rating, because it was felt that the site characteristics of each facility, 

including location of facilities, the size of the adjacent channel, etc., necessitated a site-specific risk 

assessment at each facility. The infrastructure subwatershed sensitivity ratings do provide a generalized, 



relative assessment of risk for recreation facilities. Results of the infrastructure rating (with the location of 

campgrounds within 300 ft of channels) are shown in Figure 6. 

Water Uses 

Water from the forest supports domestic, livestock, wildlife and fish, recreational, and agricultural uses 

downstream, and all watersheds within the analysis area are highly valued for this reason. Additionally, 

water for domestic use is captured by and delivered from the C.C. Cragen Reservoir. Substantial surface 

water is stored close to its source in stockponds or tanks, where it used for stock water and wildlife 

purposes. Numerous agricultural diversions exist on the lower reaches of Oak, Beaver, and West Clear 

Creeks and the Verde River. 

Ratings of relative subwatershed values for water uses were based on a combination of all these factors. 

The amount of water (acre ft) diverted in each watershed was determined, and subwatersheds with no 

diversions were given a low value, watersheds with less than 500 acre ft diverted (annually) were classed 

as moderate, and those with greater than 500 acre ft were rated as high. GIS was used to obtain a count of 

tanks per subwatershed. Subwatersheds were divided into three classes: those subwatersheds with 16 or 

fewer tanks were given the lowest value, those with 17 to 32 had moderate value, and those with more 

than 32 received the highest rating. Tanks and diversions were given equal weight, and were combined to 

produce a single water resource score. These values were then divided into thirds, with the highest third 

of subwatersheds given a rating of “high.” Finally, all subwatersheds that contribute flow to the C.C. 

Cragen reservoir were rated as high. The results of the water-uses rating are displayed in Figure 7. 

Riparian and Spring Habitats 

Relative to other areas of the country, the amount of aquatic and riparian habitat (including springs) on 

the CNF is limited. Riparian areas represent 0.7% of the area on the Forest. These spatially limited areas 

provide habitat for 80% of the Forest’s bird species, including neotropical species. Eighty percent of the 

Forest’s vertebrate species depend on riparian habitat for at least half of their life cycles. These habitats 

are vitally important as habitat for numerous reptiles and amphibians not listed above and other aquatic 

organisms, such as macroinvertebrates. Springs also provide habitat for aquatic and riparian species, 

including numerous endemic macroinvertebrate species. 

The relative value of subwatersheds for this resource was based on two data sources: miles of riparian 

habitat and the number of springs. GIS was used to determine the miles of riparian habitat in each 

subwatershed. As with other attributes, values for each watershed were ranked and then grouped into 

thirds, with subwatersheds with the most riparian habitat (>17 miles) given the highest scores. Forest GIS 

data for springs were used to determine the number of springs per watershed; these were then grouped 

into thirds. A riparian-spring rating was obtained by combining the subwatershed scores for the individual 

factors, with the riparian value given twice as much weight as the spring rating. To be clear, ratings of 

“high” were given a score of 3, and low ratings were given a score of 2. The combined scores were then 

ranked and divided into thirds, with the highest third rated as high value. 



Figures 7 and 8. Relative Ratings of Water Uses and Riparian and Spring Habitats 



It should be noted that the ID team questioned the accuracy of the stream spring layer because it only 

includes about 200 springs and there are at least 100-150 more known springs not digitized in the forest 

GIS. Additional spring data were obtained from Northern Arizona University (NAU). NAU and other 

studies have identified at least 100-150 more springs located in the fifth-field watersheds included in this 

assessment. 

Results of the riparian spring ratings are shown in Figure 8. 

Perennial Stream Habitat 

As mentioned earlier, perennial stream habitat on the CNF is relatively uncommon, and supports a wide 

variety of environmental and human uses. Initially, streams were combined with riparian and spring 

habitat, but further consideration by the ID team resulted in the decision to look at the perennial stream 

resource by itself. The team felt that the data for perennial streams were slightly better than that for either 

riparian habitat or springs, and that the existing and future demands on the perennial stream resource 

justified the switch. Miles of stream were calculated for each subwatershed. The results, displayed in 

Figure 9, reflect ratings based on ranking of the subwatersheds by miles of stream and then grouping 

them into thirds. The break for these groupings is less than 16 miles for low, and greater than 27 miles for 

a high rating. 

Figure 9. Relative values of subwatershed for perennial stream habitat 



STEP 2 - Assess Exposure 

Background 

During initial work on this assessment, exposure was included after a generic assessment of water 

sensitivity. In the final assessment procedure, exposure was evaluated prior to sensitivity. This allowed 

the team to focus on a narrower list of potential hydrologic changes, derived from consideration of how 

predicted exposure would affect hydrology, and which of those changes were important to the water 

resource values included in the assessment. 

Historic Changes 

The first step in assessment of exposure of the selected watersheds to potential climate change was to 

look at relevant historic climatic data. Review of some available long-term data from Flagstaff shows a 

general pattern of warming (Figure 10), with a less-clear pattern relative to precipitation and snowfall 

(Figure 11). Regional long term data from the Arizona Water Atlas (Figure 12) indicates a much more 

dramatic increase in air temperature since 1960, and a decline in precipitation starting about 1966, except 

for a few years of above-average precipitation in the late 1970’s to mid 1980’s. 

Figure 10. Average daily air temperatures from Flagstaff, 1950-2006 (Staudenmaier et al. 2007) 



Figure 11. Snowfall at Flagstaff (Staudenmaier et al. 2007) 

Figure 12. Air temperature and precipitation from the Central Highlands of Arizona 1930-2005 (ADWR 2011) 



Modeled Predictions 

Available to the team were predictions of climate change prepared by the Climate Impacts Group (CIG) 

of the University of Washington. CIG compared available predictions with historic data for the western 

United States, and combined models with the best correlations to develop composite models for the 

western United States (Littell et al. 2011). Downscaled data from these models were provided to National 

Forests participating in the WVA pilot, including the Coconino NF. This analysis used the CIG composite 

model, and predictions for 2030 and 2080. These were compared for the composite modeling of the 

historic condition. 

The models predict nearly-uniform air temperature increases across the Coconino NF, of about 4 degrees 

F in 2030, and 7 degrees F in 2080. Modeled comparisons, by season, are displayed in Table 3. Results 

for maximum July temperatures in 2030, as compared to the historic condition, are shown in Figure 13. 

Season Historic 2030 2080 

2030 

Change 

2080 

Change 

DJF 50.6 53.9 56.7 3.3 6.1 

MAM 66.7 70.7 73.8 3.9 7.1 

JJA 87.0 91.4 94.5 4.4 7.6 

SON 70.6 75.0 78.3 4.4 7.8 

Annual 68.7 72.7 75.9 4.0 7.1 

Table 3. Results from CIG composite model for air temperature. Values are averages for the entire analysis area. 

Figure 13. Results from CIG composite model projection for air temperature daily 

maximum for July. Results are the difference between the 2030 and historic simulations. 



The CIG applied the Variable Infiltration Capacity (VIC) (Liang et al. 1994) model to their modeled 

changes in temperature and precipitation, to predict changes to different hydrologic characteristics. Of 

most interest to the ID team were changes to snow, and runoff (Figures 14-15). Predictions again show 

fairly uniform changes across the forest, but with more change at higher elevations. This is logical, as this 

is where the most snow currently falls. If temperatures increase, a decrease in snow could be expected, 

with resultant changes in runoff timing and amount. 

Figures 14 and 15. Left, Predicted changes in Snow Water Equivalent (mm) between modeled historic and modeled 

conditions in 2070, based on the CIG composite model. Right, Predicted changes in runoff (mm/acre) between 

modeled historic and modeled conditions in 2030, based on the CIG composite model. 

The CIG composite model predicts almost no change in the annual precipitation, but does predict changes 

in the timing, with less precipitation falling in the spring, and more delivered by monsoons in the fall. 

Results of this modeling are shown in Table 4, and are averages for all the watersheds in the analysis area. 

Month Historic 2030 2030 2080 2080 

January 2.4 2.5 0.1 2.3 -0.2 

February 2.4 2.5 0.1 2.5 0.2 

March 2.4 2.0 -0.4 1.9 -0.5 

April 1.4 1.1 -0.3 0.9 -0.5 

May 0.6 0.4 -0.2 0.3 -0.2 

June 0.4 0.4 0.0 0.4 0.0 

July 2.3 2.3 0.1 2.8 0.6 

August 3.1 3.3 0.2 3.9 0.8 

September 1.9 2.5 0.6 2.6 0.8 

October 1.6 2.0 0.4 2.1 0.5 

November 1.6 1.5 -0.1 1.3 -0.3 

December 2.4 2.2 -0.2 2.3 -0.1 



Annual 22.5 22.8 0.4 23.4 0.9 

Table 4. Modeled precipitation (inches) and predicted changes from historic, by month for the Coconino 

NF analysis area 

The team also considered modeling conducted by Rajagupal (Rajagupal et al. 2010) in his assessment of 

hydrologic change in the Black and Verde Rivers. This analysis included the entire WVA area, with the 

exception of the Upper Clear Creek (East Clear Creek) watershed. The selection of these models was 

based on a “best fit” comparison of all available models with historic temperature and precipitation 

records that was completed by Dominguez et al. (2009). Some of their results are displayed in Figure 16, 

and show a fairly substantial decrease in spring runoff for all future projections, with a slight increase in 

fall flows. 

Figure 16. Simulated annual hydrograph for the Salt and Verde Rivers, based on VIC modeling. Periods1:2009- 

2038; 2:2039-2068; and 3:2069-2098. 

Hydrologic Changes of Concern 

The Forest team considered the potential changes as indicated by the CIG and Rajagupal modeling, and 

considered how these potential changes might impact the selected aquatic resources. The following is a 

brief summary of those considerations for each water resource value. 

Herpetiles 

• Less spring precipitation and runoff could result in drying of springs wetland habitats such that 

habitats might not persist through the summer, resulting in reduced populations or loss of species. 

• Dispersal might be improved in fall (more water). 

Warm Water Species 

• Natives spawn in spring triggered by snowmelt hydrograph, spawning success may be reduced. 



• Springs and headwaters are now important to natives due to the presence of invasives 

downstream. These habitats may be further restricted, resulting in reduced populations or loss of 

species. 

• Decrease in perennial stream habitat is likely. 

• Increased water temperatures are likely; in habitat with poor cover, temperatures could approach 

tolerance limits. 

• Reduced connectivity due to reduction in perennial (and seasonal) habitat. 

• Increase in flows in the fall could trigger spawning and might result in less overwinter survival. 

• Higher water temperatures result in lower O2 and higher primary productivity. 

Water Uses 

• Runoff will come earlier and baseflow will decrease during critical, dryer periods. 

• Less flow during periods of current diversion. 

• Warmer temperatures result in higher evaporative loss from reservoirs. 

Riparian and Stream Habitats 

• Year-round utilization of riparian vegetation by ungulates in Upper Clear and Upper West Clear 

Creek. This has led to impacts to aspen and other tree species in other areas. 

• Lower water tables will shrink the riparian areas longitudinally and by width. 

• Conversion of interrupted perennial streams to intermittent is likely. 

• Conversion of intermittent riparian areas to ephemeral or non-riparian areas is likely. 

• Reduced water quality from loss of buffer. 

• Changes to energy input (allochthunous). 

• There may be some shift in ephemerals from spring to fall. 

• Likely that fall flows will be flashier, resulting in poorer water quality. 

• Perennials streams are likely to shrink. 


• Higher-intensity storms expected; peak flows will increase. 

• More peaks may occur later in spring. 

The key hydrologic process potentially affected by climate change on the CNF is the amount and timing 

of precipitation. Aquatic and riparian habitats on the CNF are not abundant, and in many cases are already 

stressed. If precipitation were reduced, or flow regimes adversely affected by timing or increased 

temperatures, loss of the habitats would be expected. 

Secondary effects are likely to further stress aquatic systems. Evapotranspiration will likely increase as a 

result of increased seasonal temperatures and longer growing seasons. Flow regimes are likely to be 

further impacted, as a result. 

STEP 3 - Consideration of Watershed Sensitivity and Watershed Condition 

The current condition of the watersheds is important because it will affect how each watershed responds 

to changes in hydrologic processes. In this step, the existing condition of watersheds within the 

assessment area was categorized in terms of current condition and natural sensitivity to potential change. 

The assumption driving this analysis is that watersheds in good condition are more resilient than 

watersheds in poor condition. It is also assumed that resilient watersheds will respond better (change less 

in terms of outputs and ability to support resources) than watersheds that lack resiliency. 



Sensitivity of each subwatershed to change, including hydrologic changes that might result from a 

changed climate, was determined for each resource value by considering natural and anthropogenic 

factors most important in affecting the condition of these watersheds. In this exercise, the team assigned 

weightings to each factor based on professional judgment. Both stressors (factors that negatively impact 

condition) and buffers (factors that improve condition) were included. Factors for each resource, with 

their respective weights, are listed in Table 5. 

Of note is the importance of instream water rights as a buffer to possible impacts of climate change. 

Water rights are highly weighted buffers for five of the six water resource issues. Forest efforts in 

acquiring these rights substantially increase the chance of maintaining critical water resource values. 

Condition Factor 

Water Resource Issues 


Herpetiles 

Warm Water Fishes 

Streams 

Riparian/Springs 

Water Uses 


Data Source 

Riparian Vegetation 4 4 4 4 WCA 

Disease (chitrid) 4 Forest Data 

Invasive aquatic species 5 5 WCA 

Terrestrial Vegetation Condition 4 4 4 1 3 WCA 

Wells, Water Diversions, and Developments 5 4 5 5 5 Professional Judgment 

Invasive Riparian Species 2 3 3 WCA 

Wildfires (severe, within last 5 years) 3 3 3 3 5 Forest Data 

Road Proximity 3 4 4 2 Forest GIS 

Basin Size 4 Forest GIS 

Road Density 3 Forest GIS 

% Watershed Urbanized 4 WCA 

% Watershed >40% Slope 3 Forest GIS 

Regional/National Groundwater Policy (b) 3 2 3 Professional Judgment 

Instream Water Rights (b) 4 4 4 4 3 Forest Data 

Invasive Species Removal (b) 5 Professional Judgment 

Barriers (natural or constructed) (b) 4 Forest Data 

BAER Treatments (b) 3 Forest Data 

Table 5. Condition factors (with weightings) for each water resource. Factors that buffer condition are indicated by (b). 

A single score for each watershed was derived by multiplying each factor times its weight, and adding the 

sum of the stressors together. The sum of buffers, multiplied by their respective weights, was subtracted 

from the buffer sum. These values were then ranked and the highest third rated as having “high” 

sensitivity, the lowest third were placed in the “low” sensitivity class. Results of this classification are 

available at www.fs.fed.us/ccrc/wva/appendixes. An example (relative watershed sensitivity for stream 

habitat) is shown in Figure 17.


Figure 17. Relative watershed sensitivities for Stream Habitat 

Data sources for each sensitivity factor are listed in Table 5. The Watershed Condition Assessment 

provided much of these data. Other data sources were the Forest records, GIS, and professional judgment. 

To assess how the location of highly-valued resources related to watershed sensitivities, maps were 

created that combined these two factors. An example (for stream habitat) is displayed as Figure 18. 



Figure 18. Stream Habitat, relative rating of value and sensitivity 

As seen in Figure 18, the ID team decided to focus on the subwatersheds where resource values were 

highest, and sensitivity was either high or moderate. The logic for this approach was that since the factors 

that contributed to the sensitivity ratings were strongly influenced by management, sensitivity ratings 

could likely be influenced by focused management. Therefore, those areas where management might 

improve sensitivity were deemed to be highest priority, and are highlighted. Results for each resource are 

available at www.fs.fed.us/ccrc/wva/appendixes. The results for the combination of all resources and 

combined sensitivities are shown in Figure 19. 



Figure 19. Combined Values and Sensitivities 

STEP 4 - Evaluate and Categorize Vulnerability 

The final analysis step was to overlay areas with the highest exposure to potential climate change with 

areas identified as having the highest resource value and sensitivity. As discussed in the section on 

exposure, predicted temperature and precipitation changes across the Forest appear to be fairly uniform, 

with the greatest hydrologic change likely to be the result of changes in snowmelt. Based on review of 

the projections for change to runoff and snow water equivalent, and knowledge of the Forest conditions 

and runoff processes, the ID team decided that those watersheds with elevations above 6400 ft would 

probably be most susceptible to change, and could be termed most vulnerable. Subwatersheds were 

evaluated and placed into three categories as displayed in Figure 20. These are low exposure, with no area 

above 6400 ft; moderate exposure, with 10% of area above 6400 ft; and high exposure, with 90% of area 

above 6400 ft. 



Figure 20. Relative exposure to potential climate change effects, based on % of subwatershed above 6400 ft 

Once exposure was categorized, this rating was combined with the assessment of sensitivity and value, to 

produce a relative assessment of vulnerability for each resource, and for the combined resources. The 

vulnerability ratings for stream habitat and for all resources combined are displayed in Figures 21 and 22. 

Results for all resources are available at www.fs.fed.us/ccrc/wva/appendixes. 

Both examples reflect highest exposure at elevations above 6400 ft. Subwatersheds in the East Clear 

Creek drainage are consistently rated highly vulnerable, due to the combination of elevation, relatively 

high sensitivities, and high combined resource values. High values are associated with water uses (C.C. 

Cragen Reservoir) the presence of warm water fish species, and relatively high amounts of stream habitat. 

Pumphouse Wash in the Oak Creek watershed is the other subwatershed that displays the highest 

vulnerability. 



Figure 21. Areas with highest exposure, resource value, and sensitivity for stream habitat resource values 



Figure 22. Areas with highest exposure, resource value, and sensitivity for all water resource values combined 

STEP 5 - Response and Recommendations for Making WVA Useful for Managers 

The CNF sees the WVA results as a useful tool to help assess climate vulnerability of watersheds at 

various scales from landscape and sixth-level HUC or finer. The WVA should help identify watershed 

vulnerability to climate change necessary to identify and prioritize project-level proposal selection and 

management. 

Two management approaches and guidelines are recommended, which could integrate the WVA with the 

Watershed Condition Framework (WCF) and projects outside the WCF. The first is the sixth HUC WCF 

priority based management and the second is for projects not included in identified WCF sixth HUC 

priority watersheds or restoration action plans (WRAPs). 



Management Approach #1 and Guidelines for Integration of WVA and the WCF 

Findings of the WVA can be used to help prioritize sixth field HUC watersheds in the WCF. Findings of 

the WVA can be used to help identify project areas with moderate or high value and moderate or high 

sensitivity that are most vulnerable to climate change. Up to now, climate vulnerability has not been 

included in the prioritization of sixth field HUC watersheds in the WCF process. 

Guidelines: 

1. Focus on WCF priority watersheds first (top 5) and allow the WVA to inform prioritization and 

condition classification of the sixth HUCs. 

2. Reprioritize (if needed) selected priority watersheds based on results of WVA, to include climate 

vulnerability. 

3. Select only high-value or moderate-value watersheds from WVA. 

4. Consider highly and moderately sensitive HUCs before low-sensitivity HUCs. 

5. Filter to see if TES species are present in watershed and then consider prioritization. Start with 

species that are listed and have critical habitat (including spinedace, Gila chub, loach minnow 

spike dace, Chiricahua leopard frog) and/or critical and historical habitat. 

6. Look closer at the most vulnerable sixth-field HUCs that have high exposure to change in 

baseflow (based on VIC projections). 

7. Verify to see if stressor (high or moderate sensitivity) can be effectively managed to improve 

conditions, and if so, prioritize accordingly. 

8. The WRAP will identify practices that will enhance restoration in the short and long term. 

Management Approach #2 and Guidelines for Integration of WVA and Projects Outside of WCF 

Findings of the WVA can be used to help identify and prioritize project areas with moderate or high value 

and moderate or high sensitivity that are most vulnerable to climate change. Up to now, climate 

vulnerability has not been included in assessments or project identification process. 

Guidelines: 

1. Select only high value or moderate value watersheds from WVA. 

2. Consider highly and moderately sensitive HUCs before low-sensitivity HUCs. 

3. Filter to see if TES species are present in watershed and then consider prioritization. Start with 

ones that are listed and have critical habitat (including spinedace, Gila chub, loach minnow, and 

spike dace) and/or critical and historical habitat. Also consider the Chiricahua leopard frog. 

4. Look closer at the most vulnerable sixth-field HUCs that have high exposure to change in 

baseflow (based on VIC projections). Verify to see if stressor (high or moderate sensitivity) can 

be effectively managed to improve conditions, and if so, prioritize accordingly. 

5. Practices to enhance and improve resource conditions to be determined by IDT. 



Additional Management Considerations 

1. How do the results from WVA influence/modify existing Forest Priorities, Project Planning, and 

NEPA? The WVA highlights those valuable and sensitive water resources potentially most 

affected by climate change and better informs the need for change. 

2. How does the outcome from WVA affect Forest Planning? For the CNF, the WVA does not 

inform the current forest plan revision, because we are about to release our DEIS. For upcoming 

forests in revision, the WVA should inform the ecological need for change with respect to the 

most valuable and sensitive water resources as they may be affected by climate change. This may 

result in a change in short- and long-term planning direction. 

3. Completing WVA will allow the Forest to complete portions of the climate change scorecard. 

4. How do we integrate the climate change (WVA) into watershed condition classification? This is 

disclosed above through two potential management approaches and guidelines. 

5. How do we use WVA to guide the identification of priority baseline watersheds using the 

watershed restoration framework? This is disclosed above through two potential management 

approaches and guidelines. 

6. How does the outcome from WVA affect special-use authorizations (ski areas, additional snowmaking 

needs; water diversions; new reservoirs; expansion of reservoirs; grazing allotments)? 

The WVA will inform potential deficiencies in water quantity and location in the long-term 

(greater than 20-70 years). This may result in a change in short- and long-term planning direction 

and issuance of association special uses. 

7. How does the outcome from WVA road infrastructure affect water resources? For the CNF, the 

WVA highlighted road stream crossings as a stressor. A reduction of water quality may occur as 

riparian streamside management zones (buffers) decrease due to climate change. It also helps 

identify watersheds where decommissioning roads would improve water quality, because their 

location currently contributes to water quality degradation. 

8. How does the outcome from WVA affect recreation areas (location)? Riparian areas are expected 

to shrink and may cause developed and dispersed sites to locate even closer to water, thus 

impacting riparian function and water quality. However, the recent TMR decision should remove 

some of the recreation sites posing risk to water resources. Fall flows would be flashier, putting 

some recreation sites and roads at risk of flooding and damage. Site-specific analysis of these 

facilities is necessary to assess these risks. 

9. How does the outcome from WVA affect restoration priorities (e.g., remove barriers, reduce 

habitat fragmentation, restore and protect riparian areas)? The WVA provides additional 

information for assessment of the ecological need for change for the selected water resource 

values, and should assist in focused management in those watersheds. 

STEP 6 - Critique the Vulnerability Assessment 

1. Values identified in the WVA were limited to water resources, aquatic habitat, and biota, and did 

not include terrestrial bio-physical resources such as soils and upland vegetation. Therefore, fifth 



and sixth HUC watersheds without many water resources have not been assessed for climate 

vulnerability and will not inform sixth HUC WCF prioritization or projects outside of the WCF. 

2. Following the WVA process, watersheds without many water resources will have low value, even 

though climate change can significantly alter upland vegetation types. Thus, results are biased 

towards watersheds with many water resources. The process could be expanded to assess 

vulnerability of other resources to better assist management. 

3. We need to effectively present the framework so that Forest staff understand the process and do 

not have to start from scratch. It seems that the 6-step process varied somewhat between pilot 

Forests. 

4. Integration with the resource specialists was necessary to identify the resource values of concern, 

assess how potential hydrologic changes might affect the resources, and identify and weigh 

stressors and buffers. 

5. Need to be able to effectively address the time, cost, and relevance of performing a WVA to the 

leadership team to make it useful to managers. 

PROJECT TEAM 

The following team members contributed to this assessment: 

Amina Sena, CNF, hydrologist 

Mike Childs, CNF, fisheries 

Janie Agyagus, CNF, wildlife 

Ralph Martinez, Plumas NF, GIS 

Ken Roby, Lassen NF (retired) 

REFERENCES 

Arizona Department of Water Resources. 2011. Arizona Water Atlas Volume 5: Central Highlands 

Planning Area. 

http://www.azwater.gov/AzDWR/StatewidePlanning/WaterAtlas/CentralHighlands/default.htm 

Dominguez, F. and J. Cañon and J. Valdes. 2009. IPCC-AR4 climate simulations for the Southwestern 

US: the importance of future ENSO projections. Climatic Change DOI: 10.1007/s10584-009-9672-5 

Liang, X., D. P. Lettenmaier, E. F. Wood, and S. J. Burges. 1994. A Simple hydrologically Based 

Model of Land Surface Water and Energy Fluxes for GSMs, J. Geophys. Res., 99(D7), 14,415-14,428. 

Littell, J.S., M.M. Elsner, G.S. Mauger, E.R. Lutz, A.F. Hamlet, and E.P. Salathé. 2011. Hydrologic 

Change in the Northern U.S. Rockies and Pacific Northwest: Internally Consistent Projections of Future 

Climate for Resource Management. Regional Climate and Preliminary project report, USFS JVA 09-JV- 

11015600-039. Prepared by the Climate Impacts Group, University of Washington, Seattle. 

National Oceanic and Atmospheric Administration. 1994. National Climatic Data Center. Mean Total 

Precipitation - Annual vector digital data. Available on line at: Server=geodata.library.arizona.edu; 

Service=5151; Database=atlas01; User=NCDC; Version=dbo.DEFAULT 



Rajagopal S., Gupta H. V., Troch P. A., Dominguez F., Castro C. L. 2010. Climate change impacts on 

the water balance of a semi arid catchment in central Arizona using statistically downscaled climate data. 

(in preparation). 

Staudenmaier, Mike Jr., Preston, Reginald and Sorenson, Paul. 2007. Climate of Flagstaff, Arizona. 

NOAA Technical Memorandum NWS WR-273. 76 p. 

U.S. Department of Agriculture. 1994. A Federal Guide for Pilot Watershed Analysis, Version 1.2. 

Portland, OR. 202 pp 





Sawtooth National Forest 

March, 2012 

Prepared by: 

John Chatel 

Aquatics Program Manager 

Sawtooth National Forest 

Twin Falls Idaho 


Sawtooth National Forest Watershed Vulnerability Assessment, Intermountain Region (R4)) 

FOREST CONTEXT 

The unit and area assessed is the Sawtooth National Forest (SNF) and Sawtooth National Recreation Area 

within the Upper Salmon Subbasin (4th HUC) located in Idaho in the Intermountain Region (R4) of the 

USFS (Figure 1). 

Figure 1. Location of Upper Salmon Subbasin and Sawtooth National Recreation Area, where watershed 

vulnerability assessment was completed 

PARTNERS 

Trout Unlimited and Rocky Mountain Research Station 

ASSESSMENT OBJECTIVE 

The assessment objective was to determine what influence climate change may have on infrastructure and 

key aquatic species (bull trout) within the Upper Salmon basin on the Sawtooth National Recreation Area. 

SCALE OF ANALYSIS 

The scale of the analysis used in the Sawtooth National Recreation Area; Upper Salmon Subbasin 

assessment was HUC-6 (12-digit) subwatersheds. 


Sawtooth National Forest Watershed Vulnerability Assessment, Intermountain Region (R4) 


Columbia River Bull Trout 

• Threatened Species under Endangered Species Act since1998 

• Sawtooth NF Management Indicator Species 

• More specific habitat requirements than other salmonids 

• Associated with the coldest streams; upper tolerance limits appear to be 12-15°C 

• Climate change could lead to smaller and more isolated habitat patches and the loss of local 

populations in the Upper Salmon. 

• Embryos and juveniles are vulnerable to channel scour associated with the rain-on-snow events 

and winter peak flows. 


• Roads, campgrounds, water diversions, bridges, etc., with poor drainage or in riparian areas will 

be at increased risk from rain-on-snow events and winter peak flows. 

Water 

Resource Indicators 

Value 

Infrastructure Recreation Sites 

(Campgrounds) 


System Roads and 

Trails 

Private Ownership 

Projected Hydrologic 

Changes 

Rain-on-Snow Events 

Increased Winter Peak 

Flows 

Aquatics Bull Trout Rain-on-Snow Events 

Increased Winter Peak 

Flows 

Lower Summer Base 

Flows 

Increased Summer Water 

Temps 

Table 1. Water resource values, indicators, and analysis tools 

WATERSHED SENSITIVITY 


Analysis Tools Potential Impacts 

VIC – Winter 95 

(# of days in the winter in 

which flows are among the 

highest 5% for year) 

VIC – Winter 95 

VIC – MeanSummer 

(Mean flow during June 1 

to September 30) 

Stream Temperature 

Model (Summer Maximum 

Weekly Temperature) 

Flood Damage 

Egg and Juvenile 

Scour 

Habitat Reduction 

Habitat 

Fragmentation 

Watershed sensitivity includes natural risks from increased sediment, debris flows, and landslides to fish 

populations. The following factors were considered. 

Subwatershed Vulnerability - Percent of a subwatershed with sensitive land types (e.g., inherent surface 

soil erosion, sediment yield, and mass stability) (Figure 2). 

Landslide Prone Terrain – Included are areas with a tendency for rapid soil mass movements typified 

by shallow, non-cohesive soils on slopes with shallow translational planar landsliding phenomena are


controlled by shallow groundwater flow convergence. Also included are landforms with slow soil mass 

movements with deep earth-flows and rotational slumps, snow avalanche and rock fall areas (Figure 3). 

Figure 2. Subwatershed vulnerability within the Upper 

Salmon subbasin on the Sawtooth NRA. Red areas have 

high risk, yellow have moderate risk and green have 

low risk of surface erosion. 

WATERSHED CONDITION 

Watershed condition was determined using the "Matrix of Pathways and Indicators" and Bayesian belief 

networks. The "Matrix of Pathways and Indicators" has been a consultation requirement for species listed 

under the Endangered Species Act since the late 1990’s. Baseline information was already organized and 

summarized by the matrix according to important environmental parameters for each subwatershed within 

the Upper Salmon subbasin within Sawtooth NRA. This matrix was divided into six overall pathways: 

-- Water Quality -- Channel Condition and Dynamics 

-- Habitat Access -- Flow/Hydrology 

-- Habitat Elements -- Watershed Conditions 

Each of the above pathways is further broken down into watershed condition indicators (WCIs). WCIs are 

described in terms of functionality (Appropriate {FA}, At Risk {FR}, and At Unacceptable Risk {FUR}). 

The Functioning Appropriately column represents the desired condition to strive toward for each 

particular WCI. The current condition of each WCI is represented as falling within its respective 

functionality class (Figure 5). The units of measure for WCIs are generally reported in one of two ways: 

(1) quantitative metrics that have associated numeric values (e.g., “large woody debris: > 20 pieces per 

mile”); or (2) qualitative descriptions based on field reviews, professional judgment, etc. (e.g., “physical 

barriers: man-made barriers present”). The suite of relevant WCIs, considered together, encompasses the 

environmental baseline or current condition for the subwatershed and associated aquatic resources. 

Bayesian belief networks (Lee and Rieman, 1997) were used to evaluate relative differences in predicted 

physical baseline outcomes. They are appealing because their basic structure (a box-and-arrow diagram 


Figure 3. Landslide-prone terrain within the Upper 

Salmon sub-basin on the Sawtooth NRA. Red areas have 

high risk, yellow have moderate risk, and green have low 

risk of landslides


that depicts hypothesized causes, effects, and ecological interactions) can be readily modified to reflect 

new information or differences in perceptions about key relationships (Figure 4). Outcomes also are 

expressed as probabilities, so uncertainty is explicit. 

Bayesian belief networks (BBN) were constructed through a series of meetings with Boise and Sawtooth 

Forest biologists and the Rocky Mountain Research Station in 2004 to identify what baseline condition 

we believed possible when 

multiple indicators and 

pathways had certain 

functionality outcomes. 

Conceptual models (boxand-arrow 

diagrams) that 

depicted the hypothesized 

causal relationships were 

developed to show how 

each indicator resulted in 

pathway determinations and 

specific pathway outcomes 

resulted in an overall 

physical or biological 

baseline condition. Each 

BBN network variable or 

“node” was described as a set of discrete states that represented possible conditions or values, given the 

node’s definition. Arrows represent dependence or a cause-and-effect relationship between corresponding 

nodes. Conditional dependencies among nodes were represented by conditional probability tables (CPTs) 

that quantify the combined response of each node to its contributing nodes, along with the uncertainty in 

that response. The BBN was implemented in the modeling shell Netica software (Norsys Software Corp). 

Key model assumptions included: 

Figure 4. Bayesian belief network for determining overall physical condition from the 

six matrix pathways. 

• All independent variables (Parent Nodes) in each model exert some influence on the dependent 

variables (Daughter Nodes). There are no “inert” variables in the Bayesian belief networks and 

influence diagrams. 

• Some variables may exert greater influence than others. For example, large pools and substrate 

embeddedness were “weighted” more heavily than four other WCIs in the belief network 

developed for evaluating the Aquatic Habitat pathway functional rating. In other words, the 

probabilities in the relation table reflect a belief that the functional ratings for large pools and 

substrate embeddedness exert greater influence on the overall Aquatic Habitat pathway than any 

of the other four WCIs. 

• Where all independent variables (parent node are functioning appropriately, there is zero 

probability that the overall pathway/threat (daughter node) will be functioning at risk. 

Conversely, where all independent variables (parent nodes) are functioning at risk, there is zero 

probability that the overall pathway/threat (daughter node) will be functioning appropriately. 

• The probability that the overall pathway (daughter node) is functioning appropriately decreases 

incrementally with departure from the FA condition in its parent nodes. Conversely, the 

probability that the overall pathway or risk (daughter node) is functioning at unacceptable risk 

(FUR) decreases incrementally with improvement from the FUR condition in its parent nodes. 



Figure 5. Overall physical baseline condition of subwatersheds within the Upper 

Salmon subbasin on the Sawtooth NRA. Red areas have conditions “functioning 

at unacceptable risk,” yellow areas have conditions “functioning at risk,” and green 

areas have conditions “functioning appropriately.” 



Stressors that currently affect condition or may affect condition in the future 

Stressors or threats to aquatic resources were determined by 13 indicators of past and current management 

activities. These indicators included, among others, the amount of federal ownership within each 

subwatershed, number of abandoned mines, number of dispersed and developed recreation sites, route 

densities, water diversions, culvert barriers, and allotments. (Table 2). Criteria for each indicator were 

determined based on the Forest Plan (e.g., water quality and geomorphic integrity), literature (e.g., route 

densities), distribution through histograms (e.g., recreation) and professional judgment (e.g., culvert 

barriers) to determine the level of threat. 

Indicators Low Threat Moderate Threat High Threat 

Percent Federal Ownership 85-100% 50-84% 32 sites/6 th Field 

Dispersed Recreation Sites 0-8 sites/6 th Field 9-31/6 th Field >31/6 th Field 

Developed Recreation Sites 0-1 sites/6 th Field 2-7 sites/6 th Field >7 sites/6 th Field 

Route Density 

Miles of road/sq. miles of classified and 

unauthorized roads (w/in admin 

boundaries) 

< 0.7 mi/mi2 0.71-1.7 mi/mi 2 >1.7 mi/mi 2 

RCA Route Density 

Miles of road/sq. miles of classified and 

unauthorized roads (w/in admin 

boundaries) within RCAs 

< 0.7 mi/mi2 0.71-1.7 mi/mi 2 >1.7 mi/mi 2 

Landslide Prone Road Density 0.7 mi/mi 2 

Water Diversions No Diversions 1-2 diversions/6 th Field >2 sites/6 th Field 

Culvert Barriers No barriers present Culverts are partial Barriers to all life 

barriers (passable to stages (juveniles and 

adults, but barrier to 

juveniles) or complete 

barriers, but less than 

0.5 miles are blocked 

on minor tributary 

adults) 

Water Quality Integrity 

No damaged stream 20% stream 

Ratings are based on the cumulative segments; fully damaged; may not fully segments damaged; 

effects of localized physical problems— supports beneficial support beneficial uses does not fully support 

such as poorly constructed roads, 

mineral activities, failed culverts, and 

landslides—or dispersed sources such as 

areas of extensive grazing, timber 

harvest, road construction or wildfire. 

The ratings determine the streams and 

riparian water quality relative to their 

potential. 

uses 

(303d-listed) 

beneficial uses 

(TMDL developed) 



Geomorphic Integrity 

Rating determinations are based on the 

ability of subwatershed soil-hydrologic 

conditions to function as a sponge-andfilter 

system to absorb and store inputs 

of water, and on geomorphic resilience 

of streams, and riparian and wetland 

areas. Both natural and anthropogenic 

disturbances were used to estimate 

existing geomorphic conditions of each 

subwatershed. 

Subwatershed is in 

good condition, near 

or at properly 

functioning condition, 

and has low risk from 

further disturbance. 


Subwatershed is in fair 

condition, functioning 

at risk, and has 

moderate risk from 

additional disturbance. 

Allotments No allotments Sheep/Goat allotments 

and less than 25% of 6 th 

Field in Cattle/Horse 

allotment 

Equivalent Clearcut Acres 20% 

Table 2. Indicators and criteria used to determine threats to aquatic resources 

Figure 6. Bayesian belief network for determining overall 

threat level for each subwatershed on the Sawtooth NRA. 

Subwatershed is in 

poor condition, not 

properly functioning, 

and has high risk 

from additional 

disturbance. 

Greater than 25% of 

6 th Field in 

Cattle/Horse 

allotment and 

Sheep/Goat 

allotments present 

Figure 7. Composite threat rating for subwatersheds in the Upper 

Salmon subbasin on the Sawtooth NRA. Red areas have the most 

threats, yellow areas have moderate threat levels, and green areas 

have low threat levels. 

After each indicator was rated (low, moderate, or high), outcomes were entered into a Bayesian belief 

network to determine a composite threat rating for each subwatershed within the Upper Salmon subbasin 

on the Sawtooth National Recreation Area (Figure 6). Threat ratings were used later in this analysis to 

determine bull trout persistence. A key assumption in this analysis is that subwatersheds with a higher 

composite risk rating would be more at risk to the influences of climate change.


CLIMATE CHANGE EXPOSURE MODELS 

Water Temperature 

Potential effects of increased water temperatures due to climate change to bull trout were evaluated using 

a non-spatial multiple regression stream temperature model (Isaak et al. 2010). This model was created 

using an extensive, but non-random database of stream temperature measurements within the upper 

Salmon River, Upper S.F. Payette and Upper S.F. Boise subbasins on the SNF. More than 450 

temperature measurements (Hobo and Tidbit models) were used from numerous resource agencies from 

1994–2008. The majority of thermographs were placed in streams before mid-July, geo-referenced, and 

retrieved after mid-September. This sample period encompassed the warmest portion of the year when 

variation in temperatures among areas is most pronounced and influence on fish growth, behavior, and 

distribution is potentially greatest (Scarnecchia and Bergersen 1987, Royer and Minshall 1997). 

Predictor variables (i.e., geomorphic, climatic, and categorical) were used to describe spatial and temporal 

attributes associated with the stream network. Geomorphic predictors included watershed contributing 

area, elevation, and channel slope. Predictors in this category represented relatively static features of the 

river network, valley bottoms, and upstream watersheds that were hypothesized to affect stream 

temperatures. 

Interannual variation in climatically-influenced factors such as air temperature and stream flow have 

important consequences for stream temperatures. Air temperature affects stream temperature through 

sensible heat exchange near the surface of the stream and by influencing temperatures of near surface 

groundwater, which is an important component of summer flows. Stream flow determines the volume of 

water available for heating; larger flows have greater thermal capacities and are less responsive to heating 

(Hockey et al. 1982, Caissie 2006). 

Climate predictors included air temperature measurements derived from extrapolations of the observed 30 

year trends at cooperative weather stations (Ketchum and Stanley) on the Sawtooth National Forest, and 

the 50 year trends at the USGS gauges (S.F. Boise River near Featherville, S.F. Payette River at Lowman, 

and Salmon River below Yankee Fork near Clayton) with the longest records on or near the SNF. The air 

temperature data between weather stations was strongly correlated (r 2 = 0.74–0.91), so the individual time 

series were averaged and the same summary metrics that were applied to model stream temperatures were 

applied (i.e., MWMT). Flow data were obtained from two USGS stream gauges in the basin (Twin 

Springs and Featherville gauges). These two sets of data were also strongly correlated (r 2 = 0.97) and were 

averaged to calculate annual mean flow (m 3 /s) from 15 July to 15 September. 

Air temperature projections, used in the water temperature model, assume climate change will continue at 

the same rate that has occurred in the last 50 years on the forest. This likely underestimates the amount of 

change (as predicted by or some IPCC climate change scenarios). These scenarios generally predict the 

rate of air temperature change to accelerate due to increased carbon dioxide (Isaak/Wegner, pers. comm.). 

The advantage of using empirical estimates is that they're based on data from the Forest, are easy to 

understand. They provide estimates comparable to those from the IPCC scenarios for future values at 

mid-century. 

Categorical predictors included effects due to increased water temperature in lake outflows, water 

diversions, wildfires, and professional judgment. All upstream wildfires that occurred within the past 20 

years were considered. Water diversion effects on water temperatures were coded from zero (when they 

diverted less than 5% of flow) to three (when they diverted more than 30% of flow). Diversion effects on 

stream temperature were assumed to extend as far as 7 km downstream of the diversion or to a confluence 

with a larger river or stream. Finally, lakes larger than 0.1 km 2, or groups of lakes, were considered to 



have an influence on water temperatures as far as 7 km downstream or to the confluence with another 

water body. All predictors were found to be significant (p


were considered low risk, 20 to 40% were considered moderate, and greater than 40% were considered 

high risk. Changes in Winter 95 were determined by seeing how many days with the highest 5% flows 

increased from current to 2040 and 2080. Subwatersheds with less than a 0.5 day increase were 

considered low risk, 0.5 to 2 day increases were considered moderate risk, and increases greater than 2 

days were considered high risk. Risk ratings for Winter 95 and MeanSummer were provided by Seth 

Wenger, based on his recent work evaluating climate variables relative to geomorphic and land use in 

determining the distributions of bull trout and other species in the Interior Columbia River Basin (Wenger 

et al. (in press)). 

EXPOSURE RESULTS 

Winter Peak Flows (Winter 95) – The Upper Salmon 

subbasin has many high-elevation subwatersheds and 

is surrounded by 12,000-foot snow-capped peaks of 

the White Cloud and Sawtooth Mountains. Cold, dense 

air sinking from the mountains into the valley is the 

main reason for the chilly early-morning temperatures 

that are frequently the lowest in the lower 48 states. As 

a result, mid-winter rain-on-snow events are currently 

very rare. Rain-on-snow events that do occur typically 

happen in late April to May. The high elevation terrain 

and cold winter temperatures should help to buffer 

snow packs from winter flooding. However, as air 

temperatures increase, this natural buffering capacity 

will diminish, especially in those subwatersheds where 

temperatures hover around freezing. By 2100, air 

temperatures in Idaho could increase by 5°F (with a 

range of 2-9°F) in winter and summer (EPA 1998). 

The VIC model projects that the risk from mid-winter 

peak flows triggered by rain-on-snows events will 

increase by 2080. Specifically, the highest 5% winterpeak 

flows average 0.88 days under current conditions 

(1977-1997), but increase to 2.6 days in 2040 and to 

4.44 days in 2080 on the Sawtooth NRA under the 

A1B emission scenario. Wenger et al. (in press) found 

some areas in the interior Columbia River basin within 

the 1977-1997 timeframe to have up to 8.4 days at the 

highest 5% winter peak flow. Thus, the current risk of 

mid-winter peak flows is relatively low in comparison 

to other areas. However, these risks will be increasing. 

By 2040, three (5.9%) of the 51 subwatersheds 

analyzed have less than a 0.5 day (low risk) increase in 

winter peak low from current; 34 (66.7%) have a 0.5 to 

2 day (moderate risk) increase from current; and 14 

(27.4%) have a greater-than-2-day (high risk) increase 

from current (Figure 8). Meadow, Stanley Lake, and 

Smiley Creek have the highest risk with each having 

over a 4 day increase in winter peak flows by 2040. 


Figure 8. Winter peak flow risk in 2040; highest 

(red); moderate (yellow); and lowest (green) 

Figure 9. Winter peak flow risk in 2080; highest 

(red); moderate (yellow); and lowest (green)


By 2080, only one (2.0%) subwatershed remains in a low risk category and three (5.9%) subwatersheds 

remain in a moderate risk category (Figure 9). The remaining 47 (92.1%) subwatersheds are in a high risk 

category, with Elk, Meadow, Pettit Lake, Smiley, Stanley Lake, and Upper Redfish Lake Creeks showing 

over a 5-day increase in winter peak flows. 

Summer Baseflows (Mean Summer) 

Figure 10. Summer baseflow risk in 2040; moderate 

(yellow); and lowest (green) 


The VIC model projects that summer baseflows 

may decrease from current conditions (1977-1997) 

by 22% in 2040 and 29% in 2080, for the entire 

Sawtooth NRA under the A1B emission scenario. 

This is not unexpected, because air temperatures 

and evapotranspiration are expected to increase. 

Increasing winter air temperatures will reduce the 

amount of snow (e.g., more precipitation falling as 

rain than snow), as already observed in several 

parts of the western United States (Aguado et al. 

1992; Dettinger and Cayan 1995). Higher spring 

temperatures will also initiate earlier runoff and 

peak streamflows in snowmelt-dominated basins 

(Aguado et al. 1992; Cayan et al. 2001). 

By 2040, 18 (35.3%) of the 51 subwatersheds 

analyzed are predicted to see moderate risks (20- 

40%) from decreases in baseflow, and 33 

subwatersheds will see low risk (< 20%) (Figure 10). 

By 2080, only 5 (9.8%) subwatersheds are predicted 

to remain in a low risk category and 42 (89.4%) 

subwatersheds in a moderate risk category (Figure 

11). The remaining 4 (7.8%) subwatersheds (Beaver, 

Elk, Fishhook, and Park-Hanna) are predicted to be 

in a high risk category, with baseflow decreases of 

37% or greater. These model predictions, however, 

should not be viewed as absolute changes, but 

instead as more reflective of a general trend of 

declining baseflows. This is because the VIC does 

not model groundwater, which causes it to 

underestimate summer flows where groundwater 

contributes. Conversely, the model also 

overestimates summer flows in drainages that lose 

stream flow. 

Figure 11. Summer baseflow risk in 2080; highest 

(red); moderate (yellow); and lowest (green) 

Still, the prediction of lower baseflows is consistent with other studies. Since 1950, stream discharge in 

both the Colorado and Columbia River basins has decreased (Walter et al. 2004). Regonda et al. (2005)


and Stewart et al. (2005) found that stream runoff steadily advanced during the latter half of the twentieth 

century and now occurs 1 to 3 weeks earlier, due largely to concurrent decreases in snowpack and earlier 

spring melt (Mote et al. 2005). These changes diminished recharge of subsurface aquifers that support 

summer baseflows (Hamlet et al. 2005). Luce and Holden (2009) found that three-fourths of the 43 gauge 

records they examined from the Pacific Northwest exhibited statistically significant declines in summer 

low flows. Luce and Holden (2009) also found that the driest 25% of years are becoming drier across the 

majority of the Pacific Northwest sites, with most streams showing decreases exceeding 29% and some 

showing decreases approaching 50% between 1948 and 2006. Sites on or near the Sawtooth National 

Forest showed similar declines in mean annual flow (Table 3). 

Site Name 

13139510 

13186000 

13302500 

Big Wood River at 

Hailey 

SF Boise River NF 

Featherville 

Salmon River at 

Salmon 

Table 3. Mean annual flow from 1948-2006 

Average 

Annual Flow 

(mm) 

25 th 

Percentile 

Change 


Median 

Change 

75 th Percentile 

Change 

Mean 

Change 

257 -31% -13% -6% -7% 

411 -43% -30% 1% -21% 

182 -42% -29% -11% -26% 

In the upper Salmon River drainage, there are numerous irrigation diversions on federal and private land 

within the Sawtooth NRA. There are nine subwatersheds (Champion, Elk, Fisher, Huckleberry, Iron- 

Goat, Park-Hanna, Pole, Slate, and Smiley Creeks) at risk from declining baseflows and water diversion 

(Table 4). Future decreases in summer baseflows in these subwatersheds are likely to have severe 

consequences for aquatic ecosystems where there are already high water demands from diversions. 

HUC-6 Name 

% Decrease in MeanSummer 

Baseflows from Current 

2040 2080 Overall Influence 


Miles of Stream 

Impacted 

Alturas Lake 24 39 None -- 

Beaver Creek 30 42 Low 1.21 

Beaver-Peach 14 25 Low 3.88 

Big Boulder Creek 109 15 Low 0.58 

Big Casino Creek 10 14 Moderate 1.06 

Big Lake Creek 11 20 None -- 

Bluett-Baker 10 22 Low 3.96 

Boundary-Cleveland 15 28 Low 5.25 

Cabin-Vat 22 34 Low 1.99 

Champion Creek 17 33 Mod/High 3.13 

East Basin-Kelly 24 30 None -- 

Elk Creek 25 53 Moderate 0.30 

Fisher Creek 24 27 High 1.95 

Fishhook Creek 25 37 None -- 

Fourth of July Creek 122 21 Low/Mod 4.52 

French-Spring 13 24 Low 5.16 

Germania Creek 13 27 None --


HUC-6 Name 

% Decrease in MeanSummer 

Baseflows from Current 

2040 2080 Overall Influence 



Miles of Stream 

Impacted 

Gold-Williams 16 29 Low/Mod 10.77 

Harden-Rough 15 28 None -- 

Hell Roaring-Mays 20 32 Low 2.63 

Holman-Mill 14 24 Low 2.88 

Huckleberry Creek 18 23 High 1.64 

Iron-Goat 26 35 High 13.22 

Joes-Little Casino 15 28 Low 5.58 

Little Boulder Creek 8 22 Low 0.15 

Lower Yankee Fork 17 23 None -- 

Meadow Creek 34 38 None -- 

Muley-Elk 15 26 None -- 

Nip and Tuck-Sunny 15 29 Low 7.27 

Park-Hanna 32 42 High 8.46 

Pettit Lake Creek 18 31 None -- 

Pole Creek 26 37 High 3.19 

Prospect-Robinson Bar 12 27 None -- 

Redfish-Little Redfish 11 29 None -- 

Sawtooth City-Frenchman 29 35 Low 3.14 

Slate Creek 15 27 Moderate 6.42 

Smiley Creek 31 35 Moderate 0.92 

Spud-Clayton 12 22 None -- 

Stanley Creek 25 34 None -- 

Stanley Lake Creek 25 33 Low 1.19 

Sullivan-Clayton 12 22 None -- 

Swimm-Martin 9 29 None -- 

Thompson Creek 5 10 None -- 

Upper EF Salmon 23 32 None -- 

Upper Redfish Lake Creek 5 33 None -- 

Upper Salmon River 31 37 Low 4.04 

Upper Warm Spring Creek 1 29 None -- 

Warm-Taylor 38 35 Low 9.00 

West Pass Creek 8 12 Moderate 0.54 

Wickiup-Sheep 11 23 Low 4.66 

Yellow Belly Lake Creek 9 31 None -- 

Table 4. Comparison of summer baseflow changes and subwatersheds with water diversions 

* Green shaded (low risk) = < 20% decrease in baseflow; Yellow shaded (moderate risk) = 20 to 40% decrease; and 

Orange shaded (high risk) = > 40% decrease in baseflow. 

* Overall water diversion influence takes into account the number of diversions and miles of stream impacted by 

water withdrawals within each subwatershed.


Summer Water Temperatures (Maximum weekly maximum temperature) 

The temperature model predicts that summer maximum weekly maximum water temperatures will see a 

steady increase over the next 70 years (0.9 o C in 2033, 1.1 o C in 2040, 1.7 o C in 2058, and 2.5 o C in 2080) 

on the Sawtooth NRA. As a result, bull trout habitat within the 15 o C optimal temperature range will see a 

steady decrease. The stream temperature model currently projects that 102 miles of bull trout habitat 

within optimal temperatures exist across the Sawtooth NRA. Suitable habitat will see a slight decrease to 

100 miles by 2040, but a substantial decrease (35%) to 66.7 miles by 2080 (Figures 12 and 13). 

Figure 12. Available thermal bull trout habitat in 

Valley Creek on the Sawtooth NRA in 2008. 

Streams with optimal temperatures are portrayed in 

purple and those outside optimal range in red. 


Figure 13. Available thermal bull trout habitat in 

Valley Creek on the Sawtooth NRA in 2080. 

Streams with optimal temperatures are portrayed in 

purple and those outside optimal range in red.


Water temperature increases are not surprising, as mean air temperatures have seen a 0.49 o C increase per 

decade (1979-2008) at local weather stations and projections show air temperatures increases of another 

3.9 o C by 2080. At the same time, annual stream flows have decreased 5%/decade (1957-2008) at local 

USGS gauging stations and are projected to decrease an additional 54% by 2080. However, not every 

future year is expected to see warmer air temperatures and lower stream. The most pronounced changes 

will likely be associated with short-term cycles such as the Pacific Decadal Oscillation and the El Nino- 

Southern Oscillation. As climate change progresses, long-term warming trends will result in more 

frequent droughts and periods of unusually warm weather that were considered extreme in the twentieth 

century. When these events occur, the most affected watersheds will be those that have a high percentage 

of low-elevations terrain and channel conditions prone to heating (wide, shallow, lack of riparian 

vegetation) (Crozier and Zabel 2006). 

Projected decreases in thermally optimal bull trout habitat are similar to those by O’Neal (2002), who 

concluded that 2%-7% of current trout habitat in the Pacific Northwest would be unsuitable by 2030, 5%- 

20% by 2060, and 8%-33% by 2090. Williams et al. (2009) also concluded that cold-water fish habitat in 

the Rocky Mountain region could lose up to 35% of its habitat by 2050 and 50% by 2100. 

Ecological Departure 

Bayesian belief networks were used to determine the overall influence of stream temperature, summer 

baseflow, and winter peak flow changes due to climate change on current and historic bull trout habitat 

(Figure 14). BBN’s were constructed through a series of meetings with Sawtooth National Forest and the 

Rocky Mountain Research Station in 2010 to determine how much collective change would need to occur 

before a certain level of ecological departure impacted aquatic habitat within each subwatershed that 

supported current or historic bull trout populations. 

Bayesian models predicted that habitat in 6 (16%) bull trout patches on the Sawtooth NRA would be at 

high risk from ecologically-departed flow and temperature conditions. It also predicted that habitat would 

be at moderate risk in 17 (46%) bull trout patches and at low risk in 14 (38%) bull trout patches. By 2080, 

risks to habitat from changed flows and water temperatures increase greatly. Only one (3%) bull trout 

patch (Big Casino Creek) would have low risk from ecologically-departed flow and temperature 

conditions, while habitat in 22 

(59%) patches would be at 

moderate risk and 14 (38%) 

patches would be at high risk. 

Bull Trout Persistence 

Bayesian belief networks were 

used to determine bull trout 

persistence in the future on the 

Sawtooth NRA. Persistence of 

bull trout was based on a 

combination of factors. These 

included (Figure 15): the influence 

of increasing stream temperature, 

decreasing summer baseflow, and more frequent winter peak flow events due to climate change; the 

composite rating for risks and threats (i.e. landslide terrain, water diversions, route density, etc.); and 

current biological (i.e., local population size, life history diversity, etc.) and physical (i.e., overall 

watershed condition) baselines. The key assumption with this approach is that smaller, weaker, bull trout 

populations will be more susceptible to climate change in patches with poor baseline conditions and with 


Figure 14. Bayesian belief network for determining ecological departure from 

changes in mean summer baseflows, winter 95 rain on snow risks, and changes in 

optimal stream temperatures for bull trout


management activities that cumulatively impact habitat annually. This assumption is supported by studies 

that found that populations in complex habitats are more stable than populations in simple ones because 

they have greater capacity to buffer the effects of environmental change (Schlosser 1982; Saunders et al. 

1990; Sedell et al. 1990; Schlosser 1991; Pearson et al. 1992). Neville et al. (2006) also showed that 

small, isolated populations were at increased risk of extinction because of demographic and genetic 

factors associated with their reduced population sizes and loss of interpopulation connectivity. 

There are, however, limitations with this approach, as follows 

1. Bull trout may persist in streams that commonly exceed their perceived thermal limits (Zoellick 

1999) because of increased availability of food, lack of competition with other species, or 

adaptations that better exploit thermal refugia or shift timing of life history transitions (Crozier 

and others 2008; Jonsson and Jonsson 2009). 

2. Baselines and management threats were assumed to remain at present levels. In reality, some 

threats will diminish due to restoration or changed management approaches, some will persist due 

to a lack of political/social will to change, and new unexpected threats will emerge. As a result, 

baseline conditions will also not stay constant. 

3. It was assumed that species and populations will continue to use and respond to the environment 

as they have in the recent past. In some instances, biological adaptation to changing environments 

could mitigate some of the challenges organisms face. 

4. Finally, there are many complex interactions between physical changes brought on by climate 

change and species’ responses to these changes. While the model is a good start, it oversimplifies 

these interactions and may inaccurately project future persistence. 

Figure 15. Bayesian belief network for determining bull trout population persistence 

Currently there are 14 patches in the Upper Salmon on the Sawtooth RNA that have reproducing bull 

trout populations. Bull trout in three of these patches are “functioning at unacceptable risk”, six patches 

are “functioning at risk,” and six are “functioning appropriately.” Populations in unacceptable or at-risk 

conditions are due to low population sizes, competition/hybridization risks with brook trout, poor habitat 

conditions, and/or moderate/high management risks. Bull trout populations in a better condition are 

characterized by relatively good habitat, larger populations, low to moderate management risks, and/or no 

brook trout present. 



Current Low Risk Populations - By 2040, 

three bull trout populations are still at low 

risk, nine populations are at moderate, and 

two are at high risk of extinction (Table 5 

and Figure 16). Two populations at low 

extinction risk (Germania and Upper Warm 

Spring Creeks) have low risk from climate 

change (i.e., frequency of winter peak flows 

averaging 1.4 days, summer baseflows 

averaging a 7% decrease, and summer water 

temperatures changing very little). The other 

low extinction risk population (Swimm- 

Martin) is projected to have moderate 

climate-change risks (i.e., frequency of 

winter peak flows averaging 2.4 days, 

summer baseflows averaging a 9% decrease, 

and summer water temperatures changing 

very little), but has good watershed that 

should give the population enough resiliency 

to withstand the predicted changes. By 2080, 

all of these populations are predicted to be subjected to a greater frequency of winter peak flows (avg. 

3.4), lower summer baseflows (avg. 28% decrease), and water temperatures outside optimal conditions 

for bull trout in lower portions of each patch. However, only the Germania population goes to a moderate 

risk of extinction from increasing effects of system roads in the headwaters and water diversions lower in 

the drainage, due to climate change. 

Current Moderate Risk Populations - Four populations (Big Boulder, Little Boulder, West Pass, and 

Fourth of July Creeks) are at moderate risk more from current and historic management impacts and 

moderate watershed conditions, than from climate change. This does not imply that there are no climate 

change impacts predicted by 2040 within 

these populations. There are still moderate 

increases in winter peak flows (avg. 0.9 days), 

and small changes to summer baseflows (avg. 

8% decrease to 15% increase) and minor 

water temperature increases. However, these 

changes are not enough to increase extinction 

risks. The remaining five bull trout 

populations (Alturas Lake, Fishhook, 

Prospect-Robinson Bar, Upper EF Salmon, 

and Wickiup-Sheep) are projected to see a 

greater frequency of winter peak flow events 

(avg. 1.6 days), less baseflow (avg. 19% 

decrease) and slightly warmer water 

temperatures that may limit the use of habitat 

Figure 17. Predicted bull trout persistence in 2080. Red 

subwatersheds are at high extinction risk; yellow are at 

moderate risk, and green are at low risk. 


Figure 16. Predicted bull trout persistence in 2040. Red 

subwatersheds are at high extinction risk; yellow are at moderate 

risk, and green are at low risk. 

during portions of the summer. By 2080 

extinction risks increase to most of the above 

bull trout populations as the frequency winter 

peak flows and summer water temperatures 

increase and summer baseflows continue to decrease (Figure 17). One additional local bull trout 

population (Wickiup-Sheep) is projected to be at high risk; nine are predicted to be at moderate risk, and 

two are predicted to be at low risk of extinction (Table 5).


Current High Risk Populations - By 2040, bull trout in Slate and Champion Creeks will be at a high 

risk of extinction, but for different reasons. Champion Creek is projected to see the loss of summer 

rearing habitat in the very lowest portion of the drainage from increased water temperatures above 15 o C 

and high risks from winter peak flows. The bull trout population is already “functioning at unacceptable 

risk” due to low densities (0.7 fish per 100m 2 ), high densities of brook trout (17.1 fish/100m 2 ), recent 

wildfire effects, and impacts to migration from irrigation diversions. Projected climate changes will likely 

increase winter peak flows enough to displace and kill newly emerged bull trout. Warmer water 

temperatures may also further decrease connectivity to migratory bull trout from the Salmon River. By 

2080, risks from winter peak flows increase further (4.4 days), water temperatures are predicted to 

increase as far as the SF Champion confluence, leaving only 2.3 miles of habitat within optimal summer 

temperatures. Furthermore, baseflows are predicted to decrease by 33%, impacting rearing habitat and 

connectivity even further, especially if irrigation demands remain constant. 

By 2040, risks to summer baseflows in Slate Creek are expected to remain low, increases to winter peak 

flows increase moderately, and summer water temperatures remain high below Silver Rule Creek, due to 

irrigation diversions. These changes result in an overall low risk from climate change. However, the bull 

trout population was still projected to be at high risk of extinction due to very low population size and 

already-poor habitat conditions from grazing, historic mining, roads, irrigation diversions, and lingering 

impacts from the 1998 Labor Day flood. Thus, by 2040, climate change will add to cumulative effects but 

will not be the main driver of extinction risks. By 2080, risks from winter peak flows greatly increase (3.7 

days), summer baseflows show a moderate decrease (27%), and summer water temperatures increase 

slightly, leaving 3.3 miles within the optimal temperature range. These risks will make it harder for an 

already-weak bull trout population to persist lower in this drainage. 

Subwatershed Name Management 

Threats 

Current 

Physical 

Condition 

Current 

Biological 

Condition 

Ecological 

Departure 


2040 2080 

Population 

Persistence 

Risk 

Ecological 

Departure 

Population 

Persistence 

Risk 

Alturas Lake Creek M FR FR M M M M 

Big Boulder Creek H FR FA L M M M 

Champion Creek M FR FUR M H M H 

Fishhook Creek M FA FR M M M M 

Fourth of July Creek M FR FR L M M M 

Germania Creek M FA FA L L M M 

Little Boulder Creek M FR FA L M H M 

Prospect-Robinson 

Bar 

M FA FA M M H M 

Slate Creek H FUR FUR L H M H 

Swimm-Martin L FA FA M L M L 

Upper EF Salmon M FR FR M M M M 

Upper Warm Spring 

Creek 

L FA FA L L M L 

West Pass Creek M FR FR L M M M 

Wickiup-Sheep H FR FR M M H H 

Table 5. Extinction risks and population persistence outcomes for bull trout-occupied subwatersheds


Overall, the predictions for bull trout do not seem promising for long-term persistence for many 

populations. The long-term climate patterns in tributary streams suggest both an expected decrease in the 

total amount of cold water stream habitat and fragmentation of some colder areas into disconnected 

“patches” of suitable habitat. Bull trout populations will likely increasingly retreat into these shrinking 

summer cold water refuges to avoid warming conditions. These restricted tributary populations may 

become more vulnerable to local extinction (Dunham et al. 1997; Dunham and Rieman 1999; Morita and 

Yamamoto 2002; Rich et al. 2003; Isaak et al. 2007). Many remaining patches will be subjected to more 

frequent winter peak flows, which will scour the streambed and destroy redds and/or kill newly emerged 

fry. Populations may also be subjected to larger, more severe wildfires (McKenzie et al. 2004; Westerling 

et al. 2006) that can remove riparian vegetation or catalyze severe channel disturbances such as debris 

flows (Luce, et al 2005). Conceivably, the combined effects of shrinking patch size and increasing 

frequency or magnitude of stream channel disturbance could chip away at what remaining resiliency these 

populations have, leaving them in a poorer condition to withstand the next series of disturbances, and 

accelerating the rate of local extinctions beyond that driven by temperature alone. 

Forest Infrastructure 

Developed recreation sites and trails within riparian conservation areas, water diversions, system roads, 

bridges, and ownership were categorized according to Forest Plan and literature criteria, histograms, and 

professional judgement, to determine the level of threat associated with each type of infrastructure (Table 

6). Bayesian belief networks were then used to evaluate the overall amount of infrastructure and risk to 

facilities within each subwatershed from winter peak flows caused by rain-on-snow events on the 

Sawtooth NRA. Those subwatersheds that have moderate/high amounts of infrastructure and high risks 

from increased winter peak flows were considered to have a high risk of damage to road and trail drainage 

and facilities within riparian areas. Subwatersheds with less infrastructure were considered to have lower 

risks from winter peak flow events. 


Low 

Threat 

Moderate High 

Percent Federal Lands 85-100% 50-84% 8 sites/6 th Field 

Water Diversions No Diversions 1-2 diversions/6 th Field >2 sites/6 th System Road Density - Miles of 

Field 

road/sq. miles (within admin 

boundaries) 

< 0.7 mi/mi 2 0.71-1.7 mi/mi 2 >1.7 mi/mi 2 

Road Stream Crossings - Number of 

road/stream crossings on perennial and 

intermittent streams based current road 

layer and NHD streams within total 

subwatershed regardless of ownership 

or administrative boundaries 

0-11 crossings crossings >23 crossings 

Bridges No bridges present 1-2 Bridges >2 Bridges 

System Trails within RCAs < 0.7 mi/mi 2 0.71-1.7 mi/mi 2 >1.7 mi/mi 2 

Table 6. Forest infrastructure and levels of risk 



Figure 18. Amount of infrastructure within the Sawtooth NRA. 

Red shaded subwatersheds have high amounts of 

infrastructure; yellow moderate amounts, and green low 

amounts. 


Forty-six subwatersheds were evaluated for 

potential impacts to infrastructure on the 

Sawtooth NRA. Of these, 19 (41%) 

subwatersheds had low amounts, 22 (48%) had 

moderate amounts, and 5 (11%) had high 

amounts of infrastructure (Figure 18). 

Subwatersheds with the most infrastructure 

included Cabin-Vat, Fourth of July Creek, Nip 

and Tuck Sunny, Pole Creek, and Slate Creek. 

As discussed, VIC projects the risk from midwinter 

peak flows triggered by rain-on-snow 

events increases substantially by 2080. 

Specifically, the highest 5% winter peak flows 

average 0.88 days under current conditions 

(1977-97), but increase to 2.6 days in 2040 and 

4.44 days in 2080 in under the A1B emission 

scenario. Currently there are 18 (39%) 

subwatersheds at low risk, 24 (52%) at 

moderate risk, and 4 (9%) from winter peak 

flows (Table 7). These numbers change 

substantially as risk from winter peak flows 

increases into the future. 

By 2080, only one (2%) subwatershed (Yellowbelly Lake Creek) continues to have a low infrastructure risk, 

while 19 (41%) subwatersheds are at moderate risk and 26 (57%) are at high risk (Table 7). 

Although the Sawtooth NRA has been actively upgrading and removing facilities from riparian areas for 

many years, these efforts may not be enough to address projected increases in winter peak flows. There are 

also substantial implications to public safety, emergency access, and impacts to aquatic ecosystems. This 

new disturbance regime may be unlike anything we have faced before and will certainly challenge the 

limited resources the Forest has to repair and move facilities. If these projected changes occur, this analysis 

will provide a road map for further assessment of subwatershed infrastructure and incremental improvement.


Overall 

Current 2040 2080 

HUC Name Infrastructure 

Amount 

Winter 95 

Risk 

Risk 

Winter 95 

Risk 

Risk 

Winter 95 

Risk 

Risk 

Alturas Lake L 0.21 L 1.77 M 4.55 M 

Beaver Creek M 0.70 M 3.43 H 5.55 H 

Beaver-Peach M 0.87 M 2.22 H 4.10 H 

Big Boulder Creek M 0.59 M 1.51 M 3.23 H 

Big Casino Creek L 4.14 M 3.69 M 5.24 M 

Big Lake Creek L 0.33 L 1.50 M 3.39 M 

Boundary-Cleveland M 0.98 M 1.32 M 1.98 M 

Cabin-Vat H 0.31 M 2.26 H 4.87 H 

Champion Creek M 0.42 L 2.72 H 4.79 H 

Elk Creek M 0.35 L 3.05 H 6.08 H 

Fisher Creek L 0.48 L 2.27 M 3.84 M 

Fishhook Creek L 1.79 M 4.18 M 6.60 M 

Fourth of July Creek H 0.38 M 1.67 H 3.53 H 

French-Spring L 0.40 L 1.63 M 3.28 M 

Germania Creek M 0.34 L 1.74 M 3.58 H 

Gold-Williams M 0.31 L 1.65 M 3.43 H 

Harden-Rough M 0.66 M 2.61 H 5.20 H 

Hell Roaring-Mays M 0.30 L 2.33 H 4.91 H 

Holman-Mill L 0.56 M 2.04 M 3.90 M 

Huckleberry Creek L 2.81 M 3.79 M 5.56 M 

Iron-Goat M 3.24 H 5.54 H 7.30 H 

Joes-Little Casino M 2.57 M 3.24 H 5.09 H 

Little Boulder Creek L 0.23 L 0.79 M 2.42 M 

Meadow Creek L 1.31 M 5.95 M 8.20 M 

Muley-Elk M 1.29 M 2.90 H 4.68 H 

Nip and Tuck-Sunny H 2.02 H 3.67 H 5.65 H 

Park-Hanna M 2.02 H 4.17 H 6.21 H 

Pettit Lake Creek M 0.22 L 2.53 H 5.45 H 

Pole Creek H 0.22 M 1.80 H 3.68 H 

Prospect-Robinson Bar L 0.77 M 2.60 M 4.28 M 

Redfish-Little Redfish L 0.90 M 4.44 M 7.29 M 

Sawtooth City-Frenchman M 0.41 L 2.79 H 5.05 H 

Slate Creek H 0.78 H 2.80 H 4.50 H 

Smiley Creek M 1.39 M 5.44 H 7.59 H 

Stanley Creek M 1.72 M 2.35 H 3.89 H 

Stanley Lake Creek M 0.89 M 5.07 H 8.04 H 

Sullivan-Clayton L 0.79 M 1.67 M 3.11 M 

Swimm-Martin L 0.93 M 3.30 M 4.96 M 

Upper EF Salmon M 0.57 M 1.72 M 3.37 H 

Upper Redfish Lake Creek L 0.28 L 2.49 M 5.32 M 

Upper Salmon River L 0.50 M 2.46 M 4.80 M 

Upper Warm Spring Creek L 0.27 L 1.56 M 3.25 M 

Warm-Taylor M 0.30 L 1.18 M 2.79 H 

West Pass Creek L 0.13 L 0.69 M 2.47 M 

Wickiup-Sheep M 0.32 L 1.19 M 2.21 H 

Yellow Belly Lake Creek L 0.01 L 0.12 M 0.38 L 

Table 7. Infrastructure risks by subwatershed from increased winter peak flows 

* Green shaded = low risk; Yellow shaded = moderate risk; and Orange shaded = high risk 



APPLICATION 

The results from this analysis can be applied to the following four main areas. 

Monitoring – Continue to expand our summer stream temperature monitoring and establish year-round 

monitoring sites in select subwatersheds that are projected to have temperature increases by 2040 and in 

higher elevation subwatersheds that are projected to have minimum temperature increase. Continue to 

monitor management activities that reduce stream shading and baseflows. Consider establishing stream 

channel/riparian monitoring sites in subwatersheds projected to see winter peak flow increases. Partner 

with other agencies and groups in these efforts. 

Watershed Aquatic Recovery Strategy (WARS) – Re-examine restoration priorities in the Forest’s 

WARS strategy to determine if designated high-priority subwatersheds should remain the focus of 

restoration. Within these and other priority subwatersheds, determine where infrastructure replacements 

or restoration can be most meaningful (i.e., improving riparian condition, streams flows, culvert barriers, 

etc.) to increase aquatic species and watershed resiliency. 

Education – Share results and develop educational tools to show how large-scale climate information can 

be used at smaller scales and what new challenges/opportunities exist. 

Improve Coordination - Forests are critical sources of water and habitat, but resource availability and 

conditions are changing, causing more uncertainty. Engage with communities and other agencies in 

adaptation strategies. 

CRITIQUE 

What important questions were not considered? – I would have liked to complete an evaluation on 

what climate will mean to fire severity and intensity in the Upper Salmon. Then see what cumulative 

impacts this would have had with other risks/threats. I would have also wanted to look at summer 

baseflow changes and water diversion closer. 

What were the most useful data sources? – By far the most important data sources for climate change 

predictions were local water temperature thermographs, weather stations, and USGS stream gauges used 

to construct the stream temperature model. The VIC model was essential for predicting changes in stream 

flow. Information on existing watershed and fish population condition and management threats was also 

critical to evaluate extinction risks to bull trout. 

What were the most important data deficiencies? – Many landscapes have some natural buffering 

capacity that will help minimize some climate change effects. We lacked information on groundwater, 

local air temperature data to determine which subwatersheds have the coldest summer temperatures, and 

water temperature data from high mountain lakes and streams that could have helped to evaluate this 

buffering capacity. 

What tools were most useful? – Bayesian belief networks were essential to evaluate the interaction of 

numerous variables and outcomes for baseline, risk/threats, ecological departure, and population 

extinction risks. Rocky Mountain Research Station stream temperature and VIC models were critical in 

looking at future climate change risks. 

What tools were most problematic? – The VIC model outputs were challenging to interpret. How much 

of an increase or decrease in flows was too much? How much change needed to occur before it would 



impact populations or destabilize watershed conditions? Without assistance from researchers it would 

have been even a more subjective process in determining risk levels from certain climate changes. 

What could have been done differently in this process? – Each pilot Forest jumped into this very 

complex topic without a clear understanding of what basic climate change data was available in their area, 

what the best models are for future climate change predictions, and how to synthesize all this information 

to answer their key questions. There is a fine line between getting too much or too little direction. Too 

much direction can stifle creative approaches, and at times it was good to struggle through what was out 

there and how best to use it. However, it would have been helpful if the steering committee had made 

contacts with key climate change researchers before Forests proceeded too far in their analysis. For 

example, where is VIC data available nationally, what scale is the data, and how should it best be used to 

answer our key questions? If I had not had assistance from Trout Unlimited and Rocky Mountain 

Research Station, it would have been very difficult to complete and interpret the VIC and stream 

temperature models. 

PROJECT TEAM 

Core Team Assistance 

John Chatel (Sawtooth NF) Charlie Luce (RMRS) 

Kerry Overton (RMRS) Bruce Rieman (Emeritus Fisheries Scientist) 

Dan Isaak (RMRS) Emily Leavitt (RMRS) 

Seth Wenger (Trout Unlimited) Dona Horan (RMRS) 

Scott Vuono (Sawtooth NF) 

Jill Kuenzi (Sawtooth NF) 


John Chatel, Aquatics Program Managers, Sawtooth National Forest 

REFERENCES 

Aguado, E., D. R. Cayan, L. G. Riddle, and M. Roos, 1992. Climatic fluctuations and the timing of 

West Coast streamflow. J. Climate, 5, 1468–1483. 

Caissie, D. 2006. The thermal regime of rivers: a review. Freshwater Biology 51:1389–1406. 

Cayan DR, Kammerdiener S, Dettinger MD, Caprio JM, Peterson DH.2001. Changes in the onset of 

spring in the western United States. Bull Am Meteorol Soc 82(3):399–415. 

Crozier, L. G., and R. W. Zabel. 2006. Climate impacts at multiple scales: evidence for differential 

population responses in juvenile Chinook salmon. Journal of Animal Ecology 75:1100–1109. 

Dunham, J. B., and B. E. Rieman. 1999. Metapopulation structure of bull trout: influences of physical, 

biotic, and geometrical landscape characteristics. Ecological Applications 9:642–655. 

Dettinger, M. D. and D. R. Cayan, 1995. Large-scale atmospheric forcing of recent trends toward early 

snowmelt runoff in California. J. of Climate, 8, 606-623. 

Dunham, J. B., G. L. Vinyard, and B. E. Rieman. 1997. Habitat fragmentation and extinction risk of 

Lahontan cutthroat trout. North American Journal of Fisheries Management 17:1126–1133. 



Elsner, M. M., L. Cuo, N. Voisin, J. S. Deems, A. F. Hamlet, J. A. Vano, K. E. B. Mickelson, S. 

Lee, and D. P. Lettenmaier. 2009. Implications of 21st Century Climate Change for the Hydrology of 

Washington State, in The Washington Climate Change Impacts Assessment: Evaluating Washington’s 

Future in a Changing Climate, edited by J. S. Littell, M. M. Elsner, L. C. W. Binder and A.K. Snover, pp. 

69-106, University of Washington Climate Impacts Group, Seattle, WA. 

Fausch, K. D., Y. Taniguchi, S. Nakano, G. D. Grossman, and C. R. Townsend. 2001. Flood 

disturbance regimes influence rainbow trout invasion success among five holarctic regions, Ecol. Appl., 

11, 1438-1455. 

Hamlet, A. F., S. Lee, K. E. B. Mickelson, and M. M. Elsner. 2009. Effects of projected climate 

change on energy supply and demand in the Pacific Northwest and Washington State, in the 

Washington Climate Change Impacts Assessment: Evaluating Washington’s Future in a 

Changing Climate, edited by J. S. Littell, M. M. Elsner, L. C. W. Binder and A.K. Snover, pp. 165-190, 

University of Washington Climate Impacts Group, Seattle, WA. 

Hamlet, A. F., P. W. Mote, M. P. Clark, and D. P. Lettenmaier. 2005. Effects of temperature 

and precipitation variability on snowpack trends in the western United States, J. Clim., 18, 4545 4561. 

Hockey, J. B., I. F. Owens, and N. J. Tapper. 1982. Empirical and theoretical models to isolate the 

effect of discharge on summer water temperatures in the Hurunui River. Journal of Hydrology (New 

Zealand) 21:1–12. 

Isaak, D.I, Luce, C.H, Rieman, B.E, Nagel, D.E., Peterson, E.E, Horan, D.L., Parkes, S., and 

Chandler, G.L. 2010. Effects of climate change and wildfire on stream temperatures and salmonid 

thermal habitat in a mountain river network. Ecological Applications, 20(5), 2010, pp. 1350–137. 

Isaak, D. J., Thurow, R. F., Rieman, B. E., Dunham, J. B. 2007. Relative roles of habitat size, 

connectivity, and quality on Chinook salmon use of spawning patches. Ecological Applications. 17: 352- 

364. 

Jonsson B and N. Jonsson. 2009. A review of the likely effects of climate change on anadromous 

Atlantic salmon Salmo salar and brown trout Salmo trutta, with particular reference to water temperature 

and flow. Journal of Fish Biology 75: 2381-2447. 

Lee, D.C. and B.E. Rieman. 1997. Population viability assessment of salmonids by using probabilistic 

networks North American Journal of Fisheries Management 17:1144-1157. 

Legendre, P. 1993. Spatial autocorrelation: Trouble or new paradigm? Ecology 74:1659–1673. 

Liang, X., D. P. Lettenmaier, E. F. Wood, and S. J. Burges. 1994. A simple hydrologically 

based model of land-surface water and energy fluxes for general-circulation models, J. Geophys. Res.- 

Atmospheres, 99, 14415-14428. 


VIC-‐2L model: Evaluation and modification, Global Planet. Change, 13, 195–206. 

Luce, C. H., and Z. A. Holden. 2009. Declining annual streamflow distributions in the Pacific Northwest 

United States, 1948–2006, Geophys. Res. Lett., 36, L16401, doi:10.1029/2009GL039407. 



Luce, Charles H., Tarboton, David G., Istanbulluoglu, Erkan, Pack, Robert T. 2005. Reply to 

comment by Jonathan J. Rhodes on ‘‘Modeling of the interactions between forest vegetation, 

disturbances, and sediment yields,’’ J. Geophys. Res., 110, F01013, doi:10.1029/2004JF000279. 

Luo, L. F., and E. F. Wood. 2007. Monitoring and predicting the 2007 U.S. drought, Geophys. 

Res. Lett., 34, 6. 

Matheussen, B., R. L. Kirschbaum, I. A. Goodman, G. M. O'Donnell, and D. P. Lettenmaier. 2000. 

Effects of land cover change on streamflow in the interior Columbia River Basin (USA and Canada), 

Hydrol. Processes, 14, 867-885. 

McKenzie, D., Gedalof, Z., Peterson, D. L., Mote, P. 2004. Climate change, wildfire, and conservation. 

Conservation Biology. 18: 890-902. 

Morita, K., and Yamamoto, S. 2002. Effect of habitat fragmentation by damming on the persistence of 

stream-dwelling charr populations. Conservation Biology. 16: 1318-1323. 

Mote, P. W., A. F. Hamlet, M. P. Clark, and D. P. Lettenmaier. 2005. Declining mountain snowpack 

in western North America, Bull. Am. Meteorol. Soc., 86, 39–49. 

Neville, H. M., J. B. Dunham, and M. M. Peacock. 2006. Landscape attributes and life history 

variability shape genetic structure of trout populations in a stream network. Landscape Ecology 21:901– 

916. 

O’Neal, K. 2002. Effects of Global Warming on Trout and Salmon in U.S. Streams, Defenders of 

Wildlife and National Resources Defense Council, Washington, DC 

Pearsons, T. N., H. W. Li, and G. A. Lamberti. 1992. Influence of habitat complexity on resistance to 

flooding and resilience of stream fish assemblages. Transactions of the American Fisheries Society 

121:427–436. 

Regonda, S. K., B. Rajagopalan, M. Clark, and J. Pitlick. 2005. Seasonal cycle shifts in 

hydroclimatology over the western United States, J. Clim., 18, 372–384. 

Rich, C. F., T. E. McMahon, B. E. Rieman, and W. L. Thompson. 2003. Local-habitat, watershed, and 

biotic features associated with bull trout occurrence in Montana streams. Transactions of the American 

Fisheries Society 132: 1053–1064. 

Royer, T. V. and G. W. Minshall. 1997. Temperature patterns in small streams following wildfire. 

Archiv fu r Hydrobiologie 140:237–242. 

Saunders, H.R., R.J. Hobbs, and C.R. Margules. 1990. Biological consequences of ecosystem 

fragmentation: a review. Conservation Biology 5:18-32. 

Scarnecchia, D. L., and E. P. Bergersen. 1987. Trout production and standing crop in Colorado’s small 

streams, as related to environmental features. North American Journal of Fisheries Management 7:315– 

330. 

Schlosser, J.J. 1991. Stream fish ecology: a landscape perspective. Bioscience 41:704-712. 

Schlosser, J.J. 1982. Trophic structure reproductive success and growth rate of fishes in a natural and 

modified headwater streams. Canadian Journal of Fisheries and Aquatic Sciences 39:968-978. 



Sedell, J.R., G.H. Reeves, F.R. Hauer, J.A. Stanford, and C.P. Hawkins. 1990. Role of refugia in 

recovery from disturbances: modern fragmented and disconnected river systems. Environmental 

Management 14:711-724. 

Stewart, I. T., D. R. Cayan, and M. D. Dettinger. 2005. Changes toward earlier streamflow timing 

across western North America, J. Clim., 18, 1136–1155. 

Walter, M.T., D.S. Wilks, J.Y. Parlange and B.L. Schneider. 2004. Increasing evapotranspiration from 

the conterminous United States. J. Hydrometeorol., 5, 405-408. 

Wenger, S. J., Luce, C. H., Hamlet, A. F., Isaak, D. J., and Neville, H. M. 2010. Macroscale 

hydrologic modeling of ecologically relevant flow metrics, Water Resour. Res., 46, W09513, doi: 

10.1029/2009WR008839. 

Wenger, S. J., Dunham, J.B., Fausch, K.D., Rieman, B.E., Luce, C. H., Young, M.K., Isaak, D. J., 

Horan, D.L., Chandler, G.L., and Neville, H. M. (in press). Role of climate and invasive species in 

structuring trout distributions in the Interior Columbia Basin. Canadian Journal of Fisheries and Aquatic 

Sciences. 

Westerling, A. L., Hidalgo, H. G., Cayan, D. R., Swetnam, T. W. 2006. Warming and earlier spring 

increases western U.S. forest wildfire activity. Science. 313: 940-943. 

Williams, J.E., Haak, A.L., Neville, H.M., and Colyer, W.T. 2009. Potential Consequences of Climate 

Change to Persistence of Cutthroat Trout Populations. N. Am. J. Fish. Manag. 29(3): 533-548. 

United States Environmental Protection Agency. 1998. Climate change and Idaho. Office of Policy, 

EPA 236-F-98-007f. 

Zoellick, B. 1999. Stream Temperatures and Elevational Distribution of Redband Trout in Idaho. Great 

Basin Naturalist 59:136– 




Shasta Trinity National Forest 

April, 2012 

Prepared by: 

Christine Mai, Forest Hydrologist 

and Fred Levitan, Steve Bachman, and William Brock 

Shasta-Trinity National Forest 

Redding, California 


Shasta Trinity National Forest Watershed Vulnerability Assessment, Pacific Southwest Region (R5) 

BACKGROUND 

Eleven National Forests across the country participated in a pilot study evaluating potential impacts of 

climate-induced hydrologic change on local water resources. Each forest identified its specific water 

resource values at risk, assessed the associated watershed sensitivities, and then considered expected 

effects from future climate change exposure to evaluate the relative vulnerabilities of forest watersheds to 

climatic change. This report summarizes the results from the Shasta Trinity National Forest, representing 

California and the Pacific South West Region. 

A primary objective of these assessments is to assist forests in developing strategies to guide forest 

management in response to climate change and promote sound resource investments. Determining areas 

that are most vulnerable to climate change impacts would help focus on the adaptation opportunities that 

may exist within these areas. Knowing what is at risk and how it may be affected presents the opportunity 

to incorporate watershed vulnerability into future management actions. Promoting resiliency in areas that 

are susceptible to hydrologic change is proposed as the appropriate management strategy. 

Water supplies, aquatic habitat, and the stability of forest infrastructure are all subject to significant 

changes as a result of climate change. More severe droughts, more frequent and larger floods, lower 

seasonal stream flows, higher peak flows, increasing water temperatures, increasing erosion and 

sedimentation are just a few of the changes that are likely to occur as a result of climate change, 

especially in the western United States. This vulnerability assessment evaluates the relative risk of impact 

from climate change to aquatic resources and infrastructure on the Shasta Trinity National Forest. 

Figure 1. Location of Shasta- 

Trinity NF, and river basin and 

climatic sections 

The Shasta Trinity National Forest manages 2.1 million acres of public land located in Northern 

California (Figure 1) with forest headquarters located in Redding California. The Forest is in the Pacific 

Southwest Region (R5) of the USFS. Mediterranean climate of northern California is characterized by hot 

dry summers and cool wet winters. All climate zones in the continental United States receive precipitation 

in the summer except California. 



Two primary ecological/climatological provinces cover the majority of the Forest; the Southern Cascade 

and the Klamath Mountain Range (Miles and Goudey 1997). There are also two river systems that drain 

the Forest (Figure 1): the Sacramento River Basin and the Trinity River, which drains into the Klamath 

Basin. 

The Southern Cascade lies on the east side of the Forest and contains the headwaters of the Sacramento 

River Basin. Elevations in the Southern Cascade range from 2,000 to 14,000 feet elevation, the range in 

precipitation is from 8 to 80 inches, with a growing season of 25 to 175 days. The Southern Cascade 

includes a number of active volcanoes, including Mount Lassen on the southern end and Mount Shasta to 

the north. 

The Klamath Mountain Province lies on the west side of the Forest and contains most of the Trinity River 

portion of the Klamath Basin as well as a the portion of the Sacramento River Basin that surrounds Shasta 

Lake. Elevations in the Klamath Province are a little lower than the Southern Cascade, ranging from 200 

to 9,000 feet elevation. Climate variability is great with precipitation ranging from 18 to 120 inches and a 

growing season of 25 to 225 days. The spectacular Trinity Alps run east-west to east along the northern 

edge of the Forest within this province. The southernmost portion of the province is the headwaters of 

California’s agricultural heartland, the Central Valley. 

Figure 2. Shasta Trinity National Forest Hydrologic units included in Watershed Vulnerability Assessment. 

HUC-4 (left), HUC-5 (center) and HUC-6 (right) were the three scales used in the analysis. 


This assessment included analysis at three scales: sub-basin (HUC-4), watersheds (HUC-5) and 

subwatersheds (HUC-6) (Figure 2). The Shasta-Trinity Watershed Vulnerability Assessment (WVA) was 

unique among the WVA pilot Forests in that multiple scales were utilized. A subbasin (HUC-4) was the 

largest assessment unit and represents the largest tributaries of the large rivers on the forest (Table 1). 

Subbasins range in size from roughly 300,000 acres to 1.6 million acres. Each subbasin is subdivided into 

watersheds (HUC-5) which range in size from roughly 40,000 acres to 200,000 acres. Watersheds are 

comprised of subwatersheds (HUC-6) which range from roughly 7000 acres to 57,000 acres. 



Klamath 

River Basin 

Subbasins Sacramento 

Subbasins 

Shasta River River Basin Lower Pitt River 

Trinity River 

McCloud River 

(Main stem) 

Sacramento Headwaters 

South Fork Trinity River Sacramento/Clear 

Cow Creek 

Cottonwood Creek 

Table 1. River Basins and nested Sub-basins on the Shasta Trinity National Forest 

The most relevant scale depends on assessment objectives and on the distribution of values and/or risks. 

Ultimately the finest scales of analysis provide the greatest level of information. If the data within the 

units are relatively equally distributed then smaller scales do not provide much additional information. 

Small scales are impractical when the scale of data available is larger than the units assessed. In this case, 

there are no differences between finer and larger scales. 

RECENT CLIMATE TRENDS IN CALIFORNIA 

Mean Summer and Winter Temperatures 

Cleland used Parameter-elevation Regressions on Independent Slopes Model (PRISM) data to analyze 

climate change across the United States. The 1961-1990 and 1991-2007 time periods were compared. The 

greatest difference in mean summer temperatures appears to be in the Southwestern United States. The 

mean summer temperatures are slightly warmer (0.6 º - 3.3ºF) throughout most of the California; 

however, in a small section in the north (home of Shasta Trinity Forest) and in a small strip along the 

Sierra Nevada, mean summer temperatures appear to be slightly cooler (0.2 º - 1.5 ºF). 

Winte (from Cleland, Summe 

2010) 


Figure 3. Winter (left) and summer (right)


Precipitation 

Differences in winter precipitation throughout California appear to have increased from 0.1 to 7.9 inches 

with the greatest increases in the north (Figures 3 and 4). California shows great variability in growing 

season precipitation, compared to the rest of the nation. Northern California received more precipitation 

(0.1 to 2.1 inches) while southern California has received less (0.1 to 1.3 inches). 

Figure 4. Winter and growing season changes in precipitation (PRISM Data: 1961-1990 vs.1991-2007) 

RECENT CLIMATE TRENDS ON FOREST (Summarized from Butz and Safford 2010) 

Mean Annual Temperatures 

Most of the Forest has had an increase of about 2 degrees Fahrenheit in mean annual temperature over the 

last 75 years, driven primarily by nighttime temperature increases (Figure 5). No changes in temperature 

have occurred at the Mt Shasta weather station (northern most portions of the forest in the Southern 

Cascade Ecoregion). PRISM data suggest mean annual temperature increases are slightly less at lower 

elevations (1ºC, 1.8º F). 



Precipitation Variability 

Figure 5. Shasta Trinity National Forest Mean Annual Temperature Trends 

Precipitation variability has significantly increased at all gauges in Sacramento River Basin (Southern 

Cascade Province) (McCloud and Mt Shasta Stations, Figure 6) on the east side of the Forest. This pattern 

in not evident in the west in the Trinity portion of Klamath Basin (Big Bar, Figure 6). 



Figure 6. Shasta Trinity National Forest Trends in Precipitation Variability 

Forest Snow Depth and Mount Shasta Glacier Trends 

Minimum and mean snow depths at all snow stations on the Forest have decreased (Figure 7). Maximum 

snow depth at all stations in the Trinity River basin has decreased over the period of record. This trend is 

not consistent across the Forest, as maximum snow depths in the Central Valley Region (the Southern 

Cascade Province, Figure 7) are increasing. Growth of glaciers on Mount Shasta is consistent with 

increase in maximum depths in the Southern Cascades (Figure 7). Shasta’s glaciers are among the few in 

the world that are still growing. Glacier changes are dictated by air temperature and precipitation. 

Warming can lead to increases in precipitation (and thus glacier ice accumulation) (Nesje et al. 2008). 



Figure 7. Trends in snow depths from snow courses on the Shasta-Trinity National Forest. Maximum, mean and 

minimum depths are shown in green, blue and red, respectively. 

Figure 8. Photographs of the Hotlum Glacier, Mount Shasta, taken September 18, 1935 (left) and August 24, 2008 

(right). Photos courtesy of Mount Shasta Climbing Rangers. 

APPROACH TO ASSESSING VULNERABILITY 

The general model used in this assessment is shown in Figure 9. The approach starts with identifying 

important aquatic resource values on the Forest that might be affected by climate change. Next, the 

potential changes to climate and the resources were assessed. The third step was to examine factors that 

might modify the response. The three components are characterized (rated and ranked) at the watershed 

scales described above. Vulnerability was derived by overlaying the products of the first three steps. 

The objective of the assessment was to provide a means of describing relative vulnerability of aquatic 

resources on the Forest to potential climate change impacts. It is important to remember the results are not 

applicable to watersheds not on the Forest, and they are not based on ecological thresholds. 



Figure 9. Model used in the Shasta-Trinity NF Watershed Vulnerability Assessment. Note that stressors are limited 

to those relative to climate change exposure. 


Three resource issues were selected for analysis, warming, drying, and extreme events. The aquatic values 

of focus are the aquatic habitats associated with lakes and streams (fish focus), ponds and springs 

(sensitive aquatic species), and infrastructure (stream crossings and near-stream recreation facilities). 

These resources are likely to be impacted by climate change in different ways. Fish populations are most 

likely to be affected by warming of rivers and streams. Sensitive aquatic species are most likely to be 

affected by the drying of ponds, small lakes, and springs. Infrastructure is at increased risk of damage 

from runoff from extreme precipitation events. 

Fisheries 

Fish species on the Forest include several USFS-sensitive species as well as species listed as threatened 

and endangered under the Endangered Species Act (ESA). ESA-listed species include Sacramento River 

winter run Chinook, Central Valley spring- and fall-run Chinook, North Coast winter coho, Northern 

California steelhead, and Great Basin Redband trout. The distribution of these species is shown in Figure 

10. Impacts to these species are likely to occur as increased temperatures reduce the amount of suitable 

habitat. California lakes have been found, on average, to be warming at 0.2 degrees per year over the past 

several decades (Schneider et al. 2009). Warmer water temperatures and shifts in timing of hydrographs 

will likely disturb breeding and rearing lifecycles, and also impact food-source organisms upon which the 

species depend, resulting in additional stress. Increased stresses could result in loss of species already at 

risk. 



Figure 10. Distribution of salmonid and resident fishes on the Shasta-Trinity NF. Density of TES fish species are 

shown for HUC-4, HUC-5 and HUC-6. 

Sensitive Species 

There are 28 USFS sensitive species on the Forest; over 70% of these are aquatic species (Table 2). Most 

of these species are already at risk due to loss of habitat and habitat fragmentation. Additional stress to 

species is probable due to influences of warming on hydrologic processes. Periods of extended drought 

would also exacerbate the effects of drying on small aquatic habitats. Timing and volume of hydrographs 

are likely to shift. These increased stresses could result in loss of habitats and the species they support. 

The non-fish species are strongly associated with springs and other water bodies less than one acre in size 

(Figure 11). This analysis uses impacts to these habitats as the proxy for species effects. 



Hardhead 

redband trout 

Fishes Amphibians & Reptiles Terrestrial & Aquatic Invertebrates 

Southern torrent 

salamander 

Foothill yellow legged 

frog 


Shasta sideband snail Shasta hesperian snail 

Wintu sideband snail 

CA floater 

(freshwater mussel) 

Steelhead Cascade frog Shasta chaparral snail Nugget Pebble Snail 

Spring-run Chinook 

salmon 

Fall-run Chinook salmon 

Shasta salamander Tehama chaparral snail Scalloped Juga (snail) 

Northwestern pond turtle 

(reptile) 

Table 2. Shasta Trinity National Forest Sensitive species (List since 2007) 

Pressley hesperian snail Montane peaclam 

Figure 11. Distribution of springs and lentic habitats less than an acre in size on the Shasta-Trinity NF. Densities of 

habitats are shown for HUC-4, HUC-5 and HUC-6. 


Forest infrastructure located in or near water bodies includes road crossings (including bridges) and nearstream 

road segments, campgrounds, and water diversion facilities. As temperatures warm and more 

energy drives the hydrologic cycle, increases in the size of peak precipitation and flow events is 

anticipated. These increases will increase the risk of damage to near channel infrastructure from increased 

winter peak flows, including rain-on-snow events. Data used to characterize location and density of 

infrastructure included the distributions of stream crossings, water diversions, and areas that are 

susceptible to debris flows, mass wasting and flooding.


CLIMATE CHANGE INFLUENCE ON HYDROLOGIC PROCESSES 

Implications of climate change on water resources are very complex. Based on climate trends already 

observed and discussion of potential effects of changing climate on hydrologic processes (Furniss et al, 

2010), the team identified several changes. These were briefly addressed in the discussion of each 

resource value, and are displayed in Figure 12. Next, the team considered how these changes might 

influence key aquatic resource values. The assessment assumes the effects will be moderated in resilient 

watersheds. These inter-relationships are shown in Table 3. 

Figure 12. Summary of likely climate change effects on hydrologic processes, and on selected resource values 

Stressors (Exposure) 

Two elements were combined to rate exposure of watersheds to climate change. The first is temperature 

increases predicted by the A2 Climate Scenario from the World Climate Research Programme's 

(WCRP's) Coupled Model Intercomparison Project phase 3 (CMIP3) multi-model dataset. This is a 

downscaled global temperature modeling output available from the University of California, Santa 

Barbara. The second element of the exposure analysis was characterization of each stream and river 

segment’s relative solar exposure. The NetMap Modeling Product (citation) was used for this 

characterization. 

Projected Temperature Increases 

The CMIP3 multi-model dataset displayed below uses an A2 emission scenario represents a world that 

has a self-reliant focus on local or regional concerns as opposed to cooperative global concerns; it’s also 

driven by greater emphasis on economics than on environmental concerns. The result is temperatures at 

the high end of the range of projections. Projected temperatures are displayed in Figure 13. Note that in 



contrast to other characterizations, temperature increases are not displayed at all three scales, because the 

downscaled data do not allow discrimination at the HUC-6 level. 

Figure 13. Projected Climate Change (World Climate Research Program (WCRP), Coupled Model Intercomparison 

Project Multi Model Dataset 

Solar Exposure 

Products from the NetMap Model (Earth Systems Institute) were utilized to display areas that have the 

greatest percentage of each hydrologic unit that is susceptible to solar exposure using digital elevation 

modeling (Figure 14). Flat areas are considered to have the greatest level of exposure, and steeper ground 

is most variable, with aspects determining overall percentages that have a higher or lower degree of solar 

exposure. 



Figure 14. Thermal exposure of streams on the Shasta-Trinity NF 

Watershed Sensitivity and Resiliency 

Numerous factors were considered in the assessment of what might modify potential changes to 

hydrologic factors. Of these, two factors were thought to be most important. These are the percentage of 

each watershed where snow is the dominate runoff process, and the percentage of each watershed 

composed of geologies where groundwater is a primary influence. 

Groundwater Influence 

Though future changes in precipitation will affect all geologies, areas with groundwater influence are less 

likely to be rapidly altered by climatic influences and should supply more reliable water sources. Because 

infiltration rates are relatively high in such areas, they buffer changes to runoff timing, and increased 

water temperature. The percentage of a hydrologic unit that contains volcanic basalt or limestone was 

used to represent areas that are ground dominated systems with limited surface water flows and a 

tempered/ delayed hydrologic response. 



Figure 15. Percentage of hydrologic units in volcanic and limestone geologies; representing groundwater influence. 

While geologies that promote infiltration and groundwater may tend to buffer climate change effects, 

areas that are currently dominated by snowmelt processes are likely to be most susceptible to change. 

Snowmelt-Dominated Hydrology 

An evaluation of the climatic subsections (Ecomap 1997) was used to rate areas most susceptible to 

hydrologic transitions based on elevation and snow dominated runoff (Figure 16). Ecological subsections 

on the Forest were ranked based on the amount of snow dominated runoff. The percentages of each 

hydrologic unit containing the ranked climatic subsections determined the overall sensitivity of the 

hydrologic units. 



Figure 16. Ranking of watershed sensitivity based on snow dominated runoff processes. Higher numerical scores 

represent higher percentage of the watershed with snow. 

Combining Values, Exposure (Stressor) and Sensitivity 

A rating of each element (resource value, exposure, and sensitivity) was derived for each watershed, and 

these scores divided into fifths to obtain relative ratings of 1-5, based solely on values on the Shasta 

Trinity National Forest. They do not represent ecological thresholds. A “one” represents the lowest value 

(or stressor). A “five” corresponds to the highest value (or stressor). 

Each of the ratings is the combination of several elements. For example, the aquatic features resource 

combined information on both springs and lakes (see Table 3). The scores were then added together using 

the weighted average approach from the WVA (USDA 2011) to obtain a total “resource value” score, a 

total “exposure” score and a total sensitivity score. 

The process of combining two data sets into one combined ranking is displayed by using both Table 3 and 

Figure 17. For example, the final “value score” in Table 3 (6 th column from the left) is multiplied by 10. 

Refer next to the matrix (Figure 17) to find the intersection of this “resource value” score (10 to 50) and 

the corresponding “exposure” score (1 to 5); this intersection (labeled from Low to High) represents the 

combined “value/exposure” ranking. Again, a “low” combined score is represented by the number 1, up 



to “high” as a 5. The process is repeated, merging this new combined data set with the sensitivity ranking. 

This is done to produce an overall score of vulnerability that includes values, stressors, and sensitivity. 

Subbasins 

Drying 

Lake 

Density 

Rank 

Spring 

Density 

Rank 

Aquatic Features Susceptible to Loss from Drying 

Values at Risk Exposure 

Sum of 

Values 

Weighted 

Value 

(Sum- 

Min/Max- 

Min) 1 

Value 

Score 

Matrix 

Value 

Score = 

Value 

x 10 

NetMap 

Thermal 

Exposure 

Rank 

2030 A2 

Global 

Climate 

Model 

Rank 


Sum of 

Exposure 

Weighted 

Exposure 

(Sum-Min 

/Max-Min) 

Exposure 

Score 

Combined 

Value & 

Exposure 

Cottonwood 1 2 3 0.1 1 10 1 1 2 0.3 2 1 

Cow 4 5 9 0.9 5 50 1 1 0.0 1 3 

Lower Pit 

River 3 2 5 0.4 

2 20 

5 5 10 1.0 5 4 

McCloud 1 4 5 0.4 2 20 4 5 9 0.9 5 4 

Sacramento 

Headwaters 2 5 7 0.6 

4 40 

2 4 6 0.5 3 4 

Sacramento/ 

Clear 4 1 5 0.4 

2 20 

1 5 6 0.5 3 2 

Shasta 5 5 10 1.0 5 50 4 4 0.8 4 5 

South Fork 

Trinity River 2 3 5 0.4 

2 20 

3 3 6 0.5 3 2 

Trinity 3 3 6 0.5 3 30 2 4 6 0.5 3 3 

Table 3. Combining multiple attributes into final scores (sample table) 

High Exposure Low 

Rank 5 4 3 2 1 Rank 

High 50 H H H MH M 50 High 

40 H H MH M ML 40 

30 MH MH M ML ML 30 

20 MH M ML L L 20 

Low 10 M ML L L L 10 Low 

Rank 5 4 3 2 1 Rank 

High Exposure Low 

Values 

Figure 17. Example of matrix used to combine resource and sensitivity (stressor) ratings. Results shown in pink 

received overall rating of “5”; those in light blue received a rating of “1”. 

Value 

Exposure 

Score 2 

It is important to note that this very simplistic model has many limitations. Other factors and more refined 

datasets could be employed to improve this model. The results presented are a first cut at identifying and 

analyzing factors that can be considered in evaluating watershed vulnerability to climate change. 

1 This calculation is based on the weighted average approach used in the Watershed Vulnerability Assessment. 

Technical Guide USDA 2011. ( 5= >0.8, 4=0.6 to 0.8, 3= 0.4 to 0.6, 2=0.2 to 0.4 and 1 =


Watershed Vulnerability Results 

Fisheries 

This assessment considered increase in water temperature considered to be the primary risk to fisheries 

and fisheries habitat. Fish values were characterized by the density of fish presence with higher weighting 

for threatened, endangered and sensitive species than for resident species. The result of the analysis for 

fish is shown in Figure 18. Areas in green contain habitats that may provide greatest resilience, and 

watersheds in red support habitats that may be the most vulnerable to impacts associated with climate 

change. Watersheds shown in yellow are considered to have moderate resilience. 

Figure 18. Combined ratings of resources, stressors and exposure produce relative ratings of watershed vulnerability 

Investing in fish habitat or watershed improvement projects is expected to be most effective in watersheds 

with high resilience (green), or moderately resilient watersheds (yellow) adjacent to watersheds with high 

resilience, because these would provide a greater level of connectivity. Enhancement of connectivity is a 

vitally important form of restoration in response to climate change. Restoration has traditionally been 

driven by a combination of political and biological considerations. It is highly important that scarce 

restoration funds for species recovery be allotted based on a hierarchy that considers resource values and 

includes long-term sustainability in the face of climate change. Site selection should prioritize areas of 

high resource value, tempered by considerations of resiliency to climate change. Areas of high resource 

value would include both population strongholds and habitat that will act as refugia from the change. 



Highest priority actions are habitat protection and improving connectivity and access to existing habitat 

not currently occupied. 

Aquatic Species 

Sensitive aquatic species are represented in the analysis by springs and lentic habitats less than one acre in 

size. The primary risks to these habitats (and the species they support) are loss of suitability from 

warming and complete loss due to drying. Resource values were characterized by the density of the small 

waterbodies. Results of this analysis are displayed in Figure 19. Areas in green are watersheds supporting 

aquatic habitats that may provide greatest resilience to impacts associated with climate. Watersheds 

depicted in red are areas where habitats may be the most vulnerable to change. Investing in sensitive 

aquatic species habitat improvement projects may be most efficient in watersheds that are most resilient, 

and in watersheds with moderate resilience (yellow) that are adjacent to more resilient watersheds. 

Developing more reliable water sources and protesting acquisition of additional water rights in may 

improve resilience in all watersheds, and may help to retain water in small ponds and springs. 

Figure 19. Vulnerability of small aquatic features to drying 




Figure 20 displays results from the assessment of relative risk to infrastructure. Resource value was based 

on relative densities of roads and recreation sites in nearstream areas, road crossings and water diversions. 

Areas depicted in green are least likely to have infrastructure affected by extreme events. Watersheds 

shown in red are expected to have the greatest changes in peak flows and will be most vulnerable to 

impacts associated with extreme events. Investing in watershed improvements that buffer runoff response 

(disconnecting road crossings, etc.) may be most efficient in watersheds with greater resilience (green). 

This model needs more work to better synthesize resource values. Wilderness areas obviously should 

have greater resiliency and lower vulnerability; at this point, trail crossings are included in the model and 

result in higher vulnerability ratings. 

Figure 20. Watershed Vulnerability to Climate Change from Extreme Events 

RESPONDING TO CLIMATE CHANGE 

In ecology, resilience describes how much disturbance a system can "absorb" without substantially 

changing its condition and structure (Bakke 2009). In regard to recovery, habitat restoration, and 

conservation of at-risk aquatic species, resiliency also requires that certain key habitat characteristics or 

processes will change little, or not at all, in response to climate change (Bakke 2009). 



It is vitally important to understand that healthy hydrologic units are the most resilient to change and thus 

are a first step in considering where to apply future management. Proven management actions that 

maintain or improve resilience include the following. 

• Maintain or increase habitat accessibility 

• Prioritize aquatic habitat connectivity in refugia 

• Road improvements to reduce sediment delivery and disconnect channel crossings 

• Implementation of erosion prevention BMPs 

• Replace undersized and damaged culverts 

• Practice water conservation practices such as replacing leaky pipes, installing floats to force 

pump shutoff, and better controlling or eliminating overflow from developed water sources. 

• Riparian improvements- thinning, enhancing native communities 

• Meadow and stream improvements 

• Maintain or increase water developments supporting key species 

• Acquire water rights for critical resources 

• Promote stricter enforcement of illegal water drafting, contest new applications for use and 

storage 

• Explore creative solutions for FERC flows, relocating species above dams, removal of natural 

barriers, collaboration and communications 

• Apply actions strategically (where infrastructure replacements or restoration can be most 

meaningful to increase aquatic species and watershed resiliency) 

The list is not complete and should be expanded to consider things like strategic planting of aquatic 

species that favor adaptation to expected change to increase survival. It could also include fuel treatment 

to break up continuity of continuous dead fuels to make the watersheds more resilient to wildfire. 

Reducing road densities and other erosion and sedimentation sources also help promote watershed 

resiliency. Maintaining or improving riparian areas through distributions of diverse native species of all 

age classes is also key. 

Maintaining and increasing habitat accessibility, accomplished primarily by replacing and removing 

anthropogenic barriers that block access to historic or suitable is also important, especially to replace 

habitat loss to warming. These actions include upgrading road stream crossings and reducing or 

mitigating the barriers associated with dams and diversions. 

The other major area of critical focus is careful management of water supplies. There is a need to consider 

potential climate change effects in the review and implementation of FERC licenses. Consider developing 

additional water sources and acquiring water rights to provide supplies for threatened and endangered 

species. Consider objecting to water-use developments that might further limit water supplies. Maintain 

and improve water infrastructure to reduce water loss and waste. Increasing the enforcement of illegal 

water drafting will become even more prevalent and more significant to maintain water in streams. Illegal 

drafting is already completely dewatering portions of streams that would otherwise be perennial. 

While the Forest has the experience and capacity to implement these actions, it does not have the 

resources to implement them everywhere. Therefore planning is needed to identify priority areas for 

implementation. Results of the vulnerability assessment should be used to review, and modify as 

necessary, existing forest improvement and restoration plans. 

Finally, there is a need to share our experience and knowledge with partners and adjacent landowners 

with whom the Forest can collaborate to provide watershed-wide climate adaptation strategies that will 



better protect our precious water resources. The Forest needs to share results and develop educational 

tools to show how large scale climate information can be used at smaller scales and what new challenges 

and opportunities exist. 

LESSONS LEARNED 

• Scale Matters 

• Simplify Assessments 

− Focus on “processes” related to key values 

− Identify, locate and prioritize solutions based on these same key processes and potential 

effects. 

• Synthesis is key and most challenging 

− Seek assistance and involve critical thinkers! 

PROJECT TEAM 

• Tyler Putt, GIS Specialist, Shasta Trinity National Forest 

• Lois Shoemaker, Fire Ecologist, Shasta Trinity National Forest 

• Ralph Martinez, GIS Specialist, Plumas National Forest 

• Brenda Olson, Biologist Fish and Wildlife Service 

• Michael Wopat, Engineering Geologist, California Geological Survey 

• Sherry Mitchell Bruker, Hydrologist, Lassen National Forest 

The above individuals provided many reference resources and participated in initial brainstorming 

processes or development of data layers and critical reviews that helped to guide this project. 

Ken Roby, Lassen National Forest and USFS Stream Systems Technology Center (retired) provided 

advice during the analysis, and edited the draft report. 

Dr Lee Benda of Earth Systems Institute provided solar exposure to stream dataset products from the Net 

Map Model. 

We acknowledge the modeling groups, the Program for Climate Model Diagnosis and Intercomparison 

(PCMDI) and the WCRP's Working Group on Coupled Modelling (WGCM) for their roles in making 

available the WCRP CMIP3 multi-model dataset. Support of this dataset is provided by the Office of 

Science, U.S. Department of Energy. 

REFERENCES 

Aguado, E., D. R. Cayan, L. G. Riddle, and M. Roos. 1992. Climatic fluctuations and the timing of 

West Coast streamflow. J. Climate, 5, 1468–1483. 

Bailey. 1994. Hierarchy of Ecoregions at a Range of Scales. 

Bakke, Jostein; Lie, Øivind; Heegaard, Einar; Dokken, Trond Martin; Haug, Gerald; Birks, 

Hilary H; Dulski, Peter; Nilsen, Trygve. 2009. Rapid oceanic and atmospheric changes during the 

Younger Dryas cold period. Nature Geoscience 2, s. 202-205. (html) 



Benda, L., Miller, D. J., Andras, K., Bigelow, P., Reeves, G., and Michael, D. 2007. NetMap: A new 

tool in support of watershed science and resource management, Forest Sci., 52, 206–219, 

Caissie, D. 2006. The thermal regime of rivers: a review. Freshwater Biology 51:1389–1406. 

Cayan DR, Kammerdiener S, Dettinger MD, Caprio JM, Peterson DH. 2001. Changes in the onset of 

spring in the western United States. Bull Am Meteorol Soc 82(3):399–415 

Crozier, L. G., and R. W. Zabel. 2006. Climate impacts at multiple scales: evidence for differential 

population responses in juvenile Chinook salmon. Journal of Animal Ecology 75:1100–1109. 

Daly, Christopher. 2010. PRISM (Parameter-elevation Regressions on Independent Slopes Model) 

climate mapping system, PRISM Climate Group director. http://prism.oregonstate.edu/ 

Dettinger, M. D. and D. R. Cayan. 1995. Large-scale atmospheric forcing of recent trends toward early 

snowmelt runoff in California. J. of Climate, 8, 606-623. 

Fausch, K. D., Y. Taniguchi, S. Nakano, G. D. Grossman, and C. R. Townsend. 2001. Flood 

disturbance regimes influence rainbow trout invasion success among five holarctic regions, Ecol. Appl., 

11, 1438-1455. 

Hamlet, A. F., P. W. Mote, M. P. Clark, and D. P. Lettenmaier. 2005. Effects of temperature 

and precipitation variability on snowpack trends in the western United States, J. Clim., 18, 4545 4561. 

Isaak, D.I, Luce, C.H, Rieman, B.E, Nagel, D.E., Peterson, E.E, Horan, D.L., Parkes, S., and 

Chandler, G.L 2010. Effects of climate change and wildfire on stream temperatures and salmonid 

thermal habitat in a mountain river network. Ecological Applications, 20(5), 2010, pp. 1350–137. 

Isaak, D. J.; Thurow, R. F.; Rieman, B. E.; Dunham, J. B. 2007. Relative roles of habitat size, 

connectivity, and quality on Chinook salmon use of spawning patches. Ecological Applications. 17: 352- 

364. 

Liang, X., D. P. Lettenmaier, E. F. Wood, and S. J. Burges. 1994. A simple hydrologically 

based model of land-surface water and energy fluxes for general-circulation models, J. Geophys. Res.- 

Atmospheres, 99, 14415-14428. 


VIC-‐2L model: Evaluation and modification, Global Planet. Change, 13, 195–206. 

Luce, C. H., and Z. A. Holden. 2009. Declining annual streamflow distributions in the Pacific Northwest 

United States, 1948–2006, Geophys. Res. Lett., 36, L16401, doi:10.1029/2009GL039407. 

Luo, L. F., and E. F. Wood. 2007. Monitoring and predicting the 2007 U.S. drought, Geophys. 

Res. Lett., 34, 6. 

Maurer, E. P., L. Brekke, T. Pruitt, and P. B. Duffy. 2007. 'Fine-resolution climate projections 

enhance regional climate change impact studies', Eos Trans. AGU, 88(47), 504. "bias-corrected and 

spatially downscaled climate projections derived from CMIP3 data and served at: http://gdodcp.ucllnl.org/downscaled_cmip3_projections/, 

described by Maurer et al (2007)." 



Matheussen, B., R. L. Kirschbaum, I. A. Goodman, G. M. O'Donnell, and D. P. Lettenmaier. 2000. 

Effects of land cover change on streamflow in the interior Columbia River Basin (USA and Canada), 

Hydrol. Processes, 14, 867-885. 

McKenzie, D., Gedalof, Z., Peterson, D. L., Mote, P. 2004. Climate change, wildfire, and conservation. 

Conservation Biology. 18: 890-902. 

Miles, Scott and Charles Goudey. 1997. Ecological Subregions of California, Section and Subsection 

Descriptions. USDA, Forest Service Pacific Southwest Region, San Francisco, CA Prepared in 

cooperation with: USDA, Natural Resources Conservation Service USDI, Bureau of Land Management. 

http://fsgeodata.fs.fed.us/other_resources/ecosubregions.html 

Mote, P. W., A. F. Hamlet, M. P. Clark, and D. P. Lettenmaier. 2005. Declining mountain snowpack 

in western North America, Bull. Am. Meteorol. Soc., 86, 39–49. 

Nesje, A. 2008. Norwegian mountain glaciers in the past, present and future. Global and Planetary 

Change 60 (2008) 10–27. Available online at www.sciencedirect.com or at 

http://www.uio.no/studier/emner/matnat/geofag/GEG2130/h09/Reading%20list/Norwegian%20mountain 

%20glaciers%20in%20the%20past,%20present%20and%20future.pdf 

Neville, H. M., J. B. Dunham, and M. M. Peacock. 2006. Landscape attributes and life history 

variability shape genetic structure of trout populations in a stream network. Landscape Ecology 21:901– 

916. 

O’Neal, K. 2002. Effects of Global Warming on Trout and Salmon in U.S. Streams, Defenders of 

Wildlife and National Resources Defense Council, Washington, DC 

Pearsons, T. N., H. W. Li, and G. A. Lamberti. 1992. Influence of habitat complexity on resistance to 

flooding and resilience of stream fish assemblages. Transactions of the American Fisheries Society 

121:427–436. 

Regonda, S. K., B. Rajagopalan, M. Clark, and J. Pitlick. 2005. Seasonal cycle shifts in 

hydroclimatology over the western United States, J. Clim., 18, 372–384. 

Royer, T. V. and G. W. Minshall. 1997. Temperature patterns in small streams following wildfire. 

Archiv fu r Hydrobiologie 140:237–242. 

Saunders, H.R., R.J. Hobbs, and C.R. Margules. 1990. Biological consequences of ecosystem 

fragmentation: a review. Conservation Biology 5:18-32. 

Schlosser, J.J. 1982. Trophic structure reproductive success and growth rate of fishes in a natural and 

modified headwater streams. Canadian Journal of Fisheries and Aquatic Sciences 39:968-978. 

Sedell, J.R., G.H. Reeves, F.R. Hauer, J.A. Stanford, and C.P. Hawkins. 1990. Role of refugia in 

recovery from disturbances: modern fragmented and disconnected river systems. Environmental 

Management 14:711-724. 

Stewart, I. T., D. R. Cayan, and M. D. Dettinger. 2005. Changes toward earlier streamflow timing 

across western North America, J. Clim., 18, 1136–1155. 



USDA Forest Service. 2011. Watershed Condition Classification Technical Guide. FS- 978. Primary 

Author John Potyndy. http://www.fs.fed.us/publications/watershed/watershed_classification_guide.pdf 

Waananen, A.O. and Crippen, J.R., 1977. Magnitude and frequency of floods in California: U.S. 

Geological Survey Water-Resources Investigations Report 77-21, 102p. 

Wenger, S. J., Luce, C. H., Hamlet, A. F., Isaak, D. J., and Neville, H. M. 2010. Macroscale 

hydrologic modeling of ecologically relevant flow metrics, Water Resources., 46, W09513, doi: 

10.1029/2009WR008839. 

Wikipedia. 2011. Climate Zones of the United States (Image) 





March, 2012 

Prepared by: 

Caty Clifton, Forest Hydrologist 

Kate Day, Hydrologist 

Allison Johnson, Fishery Biologist 


Pendleton, Oregon and Ukiah, Oregon 


Umatilla National Forest Watershed Vulnerability Assessment, Pacific Northwest Region (R6) 

BACKGROUND AND FOREST CONTEXT 

National Forests across the country are evaluating the risk posed by climate change to important water 

resources on the forests and adjoining lands. These evaluations are focused on climate-induced hydrologic 

change, impacts on water diversions and aquatic species, and interactions with infrastructure. These 

Watershed Vulnerability Assessments (WVAs) provide real world examples of issue-based and 

landscape-specific approaches to assessing the vulnerability of national forest watersheds and resources to 

climatic changes, and planning and implementing effective adaptation. 

The general intent is to display, for managers, the relative vulnerability of watersheds to climate change, 

and identify watersheds containing water “values,” or systems that may be susceptible to changes in 

hydrologic conditions (Hurd et al. 1999; Furniss et al. 2010). On the Umatilla National Forest (UNF), 

vulnerability was considered at the following two landscape and issue scales. 

1. Forestwide at the HU12 scale (162 subwatersheds have UNF ownership from


OBJECTIVES AND SCALE OF ANALYSIS 

Forestwide “Coarse Grain” Analysis 

The objective is to produce a display for resource managers showing the relative vulnerability of Forest 

watersheds to risks posed by climate change, and identify watersheds containing water “values,” 

(systems) that may be susceptible to changes in hydrologic conditions (Hurd 1999; Furniss et al. 2010). 

The analysis framework was outlined by the WVA steering committee and 12 pilot Forests with the 

overall goal of producing case studies with examples and a framework for National Forest watershed 

vulnerability assessments. 

The analysis scale was Forestwide at the subwatershed unit (12-digit hydrologic unit, or HU12). A total 

of 162 HU12 watersheds contain UNF acres; of these, 101 have 25% or more UNF acres where data and 

results are most representative. This scale was intended to provide an overview of the Forest, to 

distinguish relative vulnerability from place to place based on water resource values and non-climate 

sensitivity (resilience, condition, threats). The climate data resolution was not detailed enough for HU12level 

analysis, so data were summarized at the HU10 (watershed) scale and applied uniformly to 

subwatersheds contained within. 

Generalized Framework Steps 

Values Sensitivity Exposure Vulnerability Response 

Water Uses, 

Infrastructure, 

Aquatics 

Base Watershed 

Condition ratings, 

Resiliency factors, 

Threats 

Historic and 

Projected Climate 

(2030 and 2070) 

Winter Temperature, 

Summer 

Temperature, and 

April 1 Snow water 

equivalent (SWE) 

Focused Watersheds or “Fine Grain” Analysis for Bull Trout 


Relative rating 

based on values, 

sensitivity, and 

exposure. 

Composite and 

individual value 

ratings 

Evaluate restoration 

priorities, 

infrastructure risk, 

community 

engagement 

Our goal was to develop an understanding of climate change specific to water temperatures and suitable 

critical bull trout (Salvelinus confluentus) habitat on a HU10 forestwide scale. The analysis was focused 

within HU12 subwatersheds in the three bull trout ESU subareas on the Umatilla NF (John Day, 

Tucannon in the Snake River and Washington recovery unit, and the Umatilla - Walla Walla recovery 

unit). The aim was to delineate historic, current, and future suitable bull trout habitat using a multiple 

regression stream temperature model developed by the RMRS. 

CONNECTION TO OTHER ASSESSMENTS 

Climate change vulnerability assessments are now a component of USDA’s Strategic Plan. Region 6 has 

begun a broad-scale vulnerability assessment for multiple resources, including water uses and aquatics. 

Revision of the Blue Mountains National Forest management plans is well underway and water resource 

and aquatics issues are important aspects of planning. The Draft Forest plan identifies climate change as a 

management challenge both broadly and specifically to water resources. Two Regional aquatics strategies 

(Aquatic Restoration Strategy, 2005, and Aquatic and Riparian Conservation Strategy, 2008) do not 

explicitly address climate change implications, although results from vulnerability assessments could be


used to inform aquatic restoration and conservation emphasis and may shift priorities (location, timing, 

and restoration actions). Forest and Basin restoration strategies could be updated to incorporate results 

from this initial assessment. Resource planning efforts such as the Umatilla’s Forest Integrated vegetation 

and fire risk planning, and regulatory programs (recovery planning for listed fish, and water quality Total 

Maximum Daily Loads (TMDLs)) may also consider watershed vulnerability in a changing climate. Step 

2 of the National Watershed Condition Framework which prioritizes watersheds for restoration, could 

take into account the vulnerability of watersheds to risk posed by climate change. Other connections 

include community and regional risk assessments lead by various interest groups, including water 

managers, cities, and universities. 

COARSE SCALE ANALYSIS 

Water Resource Values 

Three categories of water resource values were evaluated, with local Forest indicators selected as most 

representative of these values: 

• Water uses – Municipal watershed, public supply watershed, Forest Service potable water 

systems, and state water rights 

• Infrastructure – Campgrounds, roads, and other developments in potentially vulnerable settings 

(within 300’ of rivers and streams mapped at 1:100K) 

• Aquatics (coarse-level) – Number of ESA listed species and Chinook salmon per subwatershed, 

and groundwater dependent ecosystem (GDE) indicators (springs, wetlands, and groundwater 

dependent streams). Fine-scale temperature analysis focused on bull trout within three ESUs. 

Resource values were classified, weighted, and summed for total composite value ratings per HU12, then 

binned into 5 value rating categories. 

Examples of water resource value attributes used in categorizing and ranking, LEFT: Water Uses, RIGHT: Aquatics 

and Infrastructure 



Sensitivity (Resiliency, Watershed Condition, and Non-climate Stressors) 

Watershed sensitivity was evaluated by combining factors representing watershed resiliency, base 

watershed condition, and non-climatic stressors. 

Resiliency Factors, or “Buffers” to climate change 

• Groundwater Dependent Ecosystems – number of springs and wetlands and overall presence of 

GDEs, including springs, wetlands, rivers, and lakes per HU12, rated (also Value indicator) 

• Watershed restoration investment – 3 categories: 0= limited or no active restoration; 1= sustained 

ongoing actions to improve conditions and habitat; and 2=Focus watersheds with Action Plans, 

more than 50% percent complete. 

Resiliency factors considered but not used in this iteration include: elevation, aspect, relief ratio, geology, 

stream density, stream type, stability (mapped landslides and stability class), and other groundwater 

indicators (meadows, permeability, faults, and alluvial deposits). 

Watershed Condition 

We used available data from the Blue Mountains Forest Plan revision watershed condition model (Gecy 

file “KWS_August2010”). Watershed condition scores (-1 to +1) from “Netweaver” decision support 

model analysis, incorporated the following factors: 

• road density, road gradient, miles in buffer as % stream mile 

• range condition (AUMs/acre, compared range use based on 2009 AUMs compared to forage 

production 

• forest vegetation as weighted departure of stand condition, and 

• aquatic habitat attributes from stream survey (LWD, pools, shade, and riparian type). 

The score is the average of upslope (roads, range, and forest vegetation) and habitat (range-riparian). 

Scores for Umatilla HUC-6 subwatersheds range from -0.5937 to +0.6473. 

Watershed Condition Rating (See figure to right) 

Watershed 

Condition 

Class 

FPR Model 

Rating 

# HUC-6 

All 

# HUC-6 

UNF 

>25% 

1. Good >0.2 22 14 

2. Fair -0.2 to 0.2 101 62 

3. Poor


Stressors or factors that may exacerbate climate change 

• Mines: coded 1= mine(s) shows no evidence of impacting water quality; 2= mine(s) has the 

potential to impact water quality; 3= mine(s) is actively impacting water quality. 

• Ditches, reservoirs: present/absent 

• Fire: percent acres burned last 10 years coded 0=0% watershed burned; 1=50% burned in the last 10 years. 

• Developments and floodplain roads: Campgrounds and developments coded 0=none,\; 1=1,\; 2>1. 

Roads coded 0=0 miles; 1=1-10 miles; 2=>10-20 miles; 3=>20 miles (also under Values). 

Overall sensitivity scoring was the simple sum of weighted factors for watershed condition, resiliency, 

and stressors, binned into 5 classes per HU12: from 1=LOW Sensitivity (High resiliency) to 5=HIGH 

Sensitivity (Low resiliency) 

We used a categorical matrix approach to combining and categorizing water resource value and 

sensitivity into “Risk-Value” groups. 

Exposure 

A growing body of published research in the Pacific Northwest shows regional trends in historic 

temperatures (warming), precipitation, declining snowpack, and streamflow (Mote 2003; Knowles et al. 

2006; Hamlet and Lettenmaier 2007). Exposure represents the pressure or change imposed by future 

climate systems outside the historic range of variability. We used University of Washington-based 

Climate Impacts Group (CIG) 

downscaled gridded data at the 

watershed scale for spatial Forest 

overlay and identification of 

locations of greatest projected 

future change. The subwatershed 

scale was considered too fine to 

apply macro-scale climatehydrologic 

model outputs (grid 

cells about 6 km 2 Historic compared to 2030 

). Changes in 

winter and summer temperatures 

range from about 3 to 5 °C 

increase but spatial differences are 

very small (


Three categorical exposure values were summed for total exposure risk (from 3, least exposure to 7, 

greatest exposure). The climate exposure risk rating for each HU6 subwatershed was then combined with 

the Risk-Value rating. 


Each step in the analysis is displayed below for the 162 HU12 subwatersheds using 5 categories for value, 

sensitivity, and exposure, and combined into a simplified three-factor “composite watershed 

vulnerability” rating. 

Watershed values were ranked 10-50 in multiples of ten in an unequal distribution of arbitrary breaks based on the 

total number of values. Rankings were based on the sum of all values categorized as follows: 3-4 values=10 (Low); 

5 values=20 (Moderate/Low); 6 values=30 (Moderate); 7-8 values=40 (Moderate/High); 9-13 values=50 (High). 

Watershed sensitivity was ranked 1-5 in an approximately equal distribution. Rankings were based on 

sum of all values categorized as follows: 5-7=1 (Low); 8=2 (Moderate/Low); 9=3 (Moderate); 10=4 

(Moderate/High); 11-14=5 (High). 



Value-Risk Matrix 

Low Risk/Value High 

Rank L ML M MH H Rank 

High 7 M MH H H H 7 High 

Vulnerability 

6 ML M MH H H 6 

5 ML ML M MH MH 5 

4 L L ML M MH 4 


Vulnerability 

Low 3 L L L ML M 3 Low 

Rank L ML M MH H Rank 

Low Risk/Value High 

Exposure was ranked from 3 to 7 and categorized as follows: 0-3=L; 4=ML; 5=M; 6=MH; 7=H. 

Data were categorized into 5 categories for values, sensitivity, and exposure, but were simplified into 3 

categories for the composite relative watershed vulnerability using the matrix. 

The composite analysis included all resource values and sensitivity and climate factors, to produce a 

composite relative watershed vulnerability rating. Two individual coarse-scale analyses were also 

performed to assess relative vulnerability of individual values for aquatic species and infrastructure in a 

similar process; however, only individual values and stressors and climate variables that could affect 

those individual values were included in the analysis. 

Individual Value Ranking: Infrastructure Vulnerability 

Infrastructure vulnerability (see figure below) was assessed using high-value developments 

(campgrounds, guard stations, and other buildings) as the value metrics. Sensitivity and vulnerability 

factors included in the analysis were similar to those used in composite analysis, with the exclusion of 

roads and developments. Change in SWE was the only climate factor used to assess exposure; changes in 

summer and winter temperature are not expected to have a direct effect on infrastructure and 

development.


Individual Value Ranking: Aquatic species vulnerability 

Aquatic species’ vulnerability was assessed using the number of focal aquatic species per subwatershed as 

the value metric. All sensitivity and threats variables, as used in the composite analysis, were used in this 

analysis. All climate factors, including winter and summer temperature and SWE, were also included in 

the analysis. Results were placed in three categories; high, medium, and low. Greatest vulnerability tends 

to be in subwatersheds with 3 focal aquatic species; however, not all subwatersheds with 3 focal aquatic 

species show high vulnerability. 



FINE SCALE ANALYSIS FOR BULL TROUT 

Bull trout (Salvelinus confluentus) was used as our aquatic focal species in the WVA because bull trout 

require cold (≤ 17 °C) and relatively low gradient, pristine waters to rear and spawn. They have a small 

thermal niche and are very responsive to changes in stream temperature. Analysis of suitable habitat on 

the UNF is necessary because bull trout are on the edge of their bioclimatic envelope (Beever et al. 2010, 

Dunham et al. 2003); the UNF is a fairly low elevation, dry forest landscape. Bull trout populations in the 

southern parts of the UNF can also be described as peripheral populations or species that are at the 

geographic edge of their range; they often have increased conservation value because they maximize 

within-species biodiversity, retain important evolutionary legacies, and may provide a “gene pool” for 

future adaptation (Haak et al. 2010). Previous research suggests future stream temperature increases on 

the forest, but influences on distribution and abundance of stream organisms is not well documented 

(Rieman et al. 2007). To begin the analysis, current bull trout distributions were identified in the 

Umatilla, Walla Walla, Tucannon, Lookingglass, and North Fork John Day (NFJD) drainages. Previous 

stream surveys conducted by USFS and ODFW/WDFW were used to verify current bull trout 

distribution. 

Multiple Regression Stream Temperature Model 

A multiple regression stream temperature model developed by the RMRS was used to model historic, 

current and future (years 2033, 2058, 2080) suitable bull trout habitat. Stream temperature model 

information and methods to the can be found at 

www.fs.fed.us/rm/boise/AWAE/projects/stream_temperature.shtml 

The regression model used observed mean weekly maximum temperature (MWMT) and physical 

parameters or predictor variables and geomorphic variables that have direct effects on stream 

temperatures. (The regression equation and coefficients can be found at 

www.fs.fed.us/ccrc/wva/appendixes.) 

Physical metrics: 

• Water diversion 

• Wildfire – Used data from the last 20 years; ~4km from the stream. 

• Groundwater Dependent Ecosystems (resiliency): number of springs and wetlands per HU12, 

rated 

Geomorphic variables or metrics (National hydrologic data set): 

• Cumulative drainage area (km 2 ) 

• Slope (%) 

• Elevation (m) 

Observed Stream Temperature and Climate Data 

Observed summer MWMT were taken from 37 locations and provided a total of 333 stream observations. 

A separate regression model was developed to predict historic and future stream temperatures using the 

same physical and geomorphic predictor vales, however, air MWMT data (1979-2009) and flow (m 3 /s) 

data (1957-2009) were considered. (Details about this regression model are available at 

www.fs.fed.us/ccrc/wva/appendixes.) 



Historic Record 2033 2058 2080 

Air MWMT (°C) 35.7 36.7 37.7 38.7 

Flow (m 3 /s) 2.02 2.00 1.98 1.96 

Air MWMT - 0.42 °C decadal increase, Flow (m 3 /s) - 0.009 m 3 /s decadal decrease 

Results: Suitable bull trout critical habitat (2033-2080) 

It was difficult to quantitatively measure suitable bull trout habitat loss between the years 2033-2080 for 

many reasons. The model predicted that only 8% of all suitable bull trout habitat forestwide would be lost 

by the year 2080 (~9,804 total miles with 769 miles lost). This underestimates loss because not all streammiles 

included in this forestwide analysis have presence of bull trout, so the calculated habitat loss seems 

small. 

When more closely examining the NFJD subwatershed, where there is known presence of migratory and 

rearing bull trout habitat, the critical habitat that is lost is approximately 22% (~81 miles of suitable 

habitat and ~18 miles lost by 2080). This may also be an underestimate because not all habitats that were 

projected “suitable” were historically or currently occupied with viable bull trout populations. From our 

current understanding, only a small percentage of streams in the upper NFJD provide rearing habitat for 

juvenile bull trout. Therefore, when looking at known juvenile bull trout distribution, a 34% loss of 

suitable bull trout habitat may be a better estimate of habitat loss in the NFJD watershed. 

Major habitat losses: 

• Tucannon - 9.43 mi 

• Mill Cr. - 20.43 mi 

• Umatilla and NF Umatilla - 15.28 mi 

• Upper NFJD watershed - 15.12 mi 

(Most of the habitat lost was tributary habitat.) 

Watersheds more resilient to bull trout habitat loss, possibly due to groundwater influence and habitat 

complexity: 

• Lookingglass 



• Little Lookingglass 

• Upper Walla Walla 

Discussion/Management Objectives 

It is important to apply this knowledge to active restoration and to highlight the importance of stream 

connectivity and aquatic organism passage. Resilience of local bull trout populations to disturbance is 

linked to the condition, structure, and interaction of populations and habitats at larger scales (Dunham and 

Rieman 1999; Neville et al. 2009; Isaak et al. 2010). Thus, active riparian restoration and improvement to 

passage barriers are important in addressing any thermal or anthropogenic barriers that may alter bull 

trout movement. In addition, because bull trout on the UNF are on the edge of their “bioclimatic 

envelope,” they may provide a leading edge for range shifts with warming temperatures, and it is 

important to establish this baseline. These peripheral populations may be our best avenue for maximizing 

future adaptive potentials for high temperature tolerance. Implementing a monitoring protocol or making 

habitat improvements to bull trout habitat can be costly and prioritizing management response is 

important, especially because this analysis shows that some watersheds have more temperature resilience 

than others. 

Prioritize Key Watersheds: Upper NFJD 

The responses of most salmonid populations to habitat alteration due to temperature increases have been 

difficult to quantify, and most efforts with bull 

trout have focused on linkages between habitat 

condition and survival of life stages. For 

example, a slow-growing resident population 

may not persist even after modest habitat 

change, while migratory or fast-growing stock 

might be viable in similar or worse situations 

(Rieman and McIntyre 1993). The bull trout 

populations in the upper NFJD and Desolation 

Creek are examples of small, isolated, slowgrowing 

populations and are especially 

vulnerable to anthropogenic disturbances such 

as road density and nonnative fish 

introductions. There have been many efforts in 

active stream restoration in the upper Granite 

Creek drainage to improve stream habitat Lookingglass Creek springs , Fall spawning survey, 2009 

complexity. Continued restoration efforts are 

essential for persistence of this bull trout population and are necessary because this population is one of 

the last strongholds on the NFJD. It is also important to mention that John Day bull trout populations have 

different allele frequencies from Walla Walla and Umatilla populations and are similar to only a few 

Grande Ronde populations (Spruell and Allendorf 1997). 

The Lookingglass drainage and the Upper Walla Walla rivers show a strong resilience to future critical 

habitat loss, possibly due to groundwater influence, few cumulative stresses (nonnative fish threats), and 

intact stream complexity. Because of these drainages, a thorough monitoring program is needed. 



SUMMARY OF FINDINGS AND MANAGEMENT ACTIONS 

Forest-scale rating of relative watershed vulnerability to climate change shows that a majority of the 

Forest has “moderate” to “high” vulnerability, using categorical indicators for Water Values, Sensitivity, 

and Exposure. Two “hot spots,” or cluster watersheds, show the highest rating: mid-Columbia marine 

influence zone (temperature vulnerability), and upper NFJD, higher elevation snow zone (water supply 

vulnerability). A total of 29 HU12 subwatersheds, or 18%, ranked highest vulnerability. (A summary of 

vulnerability factors and management options is available at www.fs.fed.us/ccrc/wva/appendixes.) 

Bull trout habitat modeling shows current habitat quality and projected losses and fragmentation in 

response to warming climate. Populations in Upper NFJD may be more susceptible to human impacts. 

Groundwater and habitat complexity may buffer climate impacts in some watersheds. More resilient areas 

in Upper Lookingglass and Walla Walla could be a focus for protection and restoration. 

Management Actions 

• Verification: Field verification of potential susceptibility to hydrologic regime changes of 

campground and other high value developments. GIS analysis of these values was limited by 

quality of spatial data; some developments may or may not be vulnerable. Field verification and 

more detailed hydrologic modeling is needed. 

• Increase resilience: Use existing programs for protecting watersheds; measures include “Best 

Management Practices”, Forest Flood Emergency Response Plan, and land allocations 

(wilderness and roadless areas as refugia). 

• Actively restore: Evaluate restoration priorities and activities, and address vulnerable 

infrastructure, passage barriers, and riparian conditions. 

• Improve coordination: Forests are critical sources of water and habitat, but resource availability 

and conditions are changing, with more uncertainty. Consider findings in Forest planning, 

Regional vulnerability assessments, and restoration strategies. Engage with communities in 

adaptation strategies. Assess current juvenile bull trout populations in the key watersheds to begin 

the process of establishing the “thermal” limit of juvenile bull trout. 

• Improve monitoring: Follow the bull trout monitoring protocol and example application in the 

Secesh River basin (published by RMRS) to design bull trout monitoring protocol for the UNF. 

• Expand inventory of culvert barriers and compile other cumulative effects that may alter bull 

trout distribution. 

• Refine modeling to address variation in stream temperature scale; for example, site versus 

systematic variation at stream, landscape, and regional scales is an issue with many temperature 

studies (Isaak et al. 2010). There is a need to collect further climatic data at finer scales and 

consult PRISM data (OSU application) to make improvements to temperature models. 



CRITIQUE 

Questions not considered: This first run-through for the Forest and initial fine-scale analysis for the bull 

trout did not address many questions, such as downstream resource values at risk. The analysis also does 

not fully represent resilience factors and did not use a full suite of climate exposure factors, including 

flow metrics. 

Most useful data sources: Forest Plan revision watershed condition data, CIG data, and Forest water 

temperature data. 

Most important data deficiencies: Physical framework, water uses data, and complexity of using 

gridded climate data. 

Useful tools: ArcGIS, RMRS temperature model, with caveats (need technical assistance) 

Problem tools: Water rights data and climate data sets. 

FUTURE WORK 

• Refine coarse-scale analysis: validate ratings, run individual values with specific climate 

exposure (Water Uses and SWE), and consider 2070 timeframe. 

• Improve fine-scale model analysis – incorporate finer-scale historic climate data into model, 

identify where habitat losses and disconnects are likely. In Forest Restoration strategy, consider 

individual actions to improve connectivity and maintain habitat. Identify “lost causes.” 

• Comparison of bull trout habitat modeling to coarse-scale aquatic species vulnerability analysis. 

• Use flow metrics in more detailed hydrologic analysis (Wenger et al, 2010). 

PROJECT TEAM 

Core Team: Caty Clifton, Forest hydrologist; Kate Day, hydrologist; Allison Johnson, fish biologist 

Support: Kristy Groves, Dave Crabtree, Tracii Hickman - fish biologists, aquatic analysis. 

Bob Gecy - watershed condition ratings from the Blue Mountains Forest plan revision, basis for 

sensitivity rating, and analysis of historic climate and gage data in the Blue Mountains 

RMRS: Dan Isaak and Dona Horan - temperature modeling and data processing assistance 

Ralph Martinez, - GIS analyst, Plumas NF - support preparing CIG climate data. 

Pilot Forests - for a community of practice; in particular, Christine Mai for risk matrix concept 

External: Ken Roby, USFS fish biologist emeritus - project support and coordination 

Rich Carmichael, ODFW - Mid Columbia Steelhead Recovery Plan vulnerability assessment example 

Climate Impacts Group: Jeremy Littell - climate data, expertise, and advice 

The Nature Conservancy, Oregon: Jenny Brown - groundwater assessment data 


Caty Clifton, Forest Hydrologist 


cclifton@fs.fed.us 

(541) 278-3822 



REFERENCES 

Beever, A., C. Ray, P. Mote, and J.L. Wilkening. 2010. Testing alternative models of climate-mediated 

extirpations. Ecological Applications 20(1): 164-178. 

Dunham, J.B., and B.E. Rieman. 1999. Metapopulation structure of bull trout: influences of physical, 

biotic, and geometrical landscape characteristics. Ecological Applications 9(2): 642-655. 

Dunham, J.B., B.E. Rieman, and G. Chandler. 2003. Influences of temperature and environmental 

variables on the distribution of bull trout within streams at the southern margin of its range. North 

American Journal of Fisheries Management 23: 894-904. 

Furniss, M.J. et al. 2010. Water, climate change, and forests: watershed stewardship for a changing 

climate (PNW-GTR-812). 

Hamlet, A.F. and D.P. Lettenmaeir. 2007. Effects of 20 th century warming and climate variability on 

flood risk in the western US (WRR v. 43). 

Hurd et al. 1999. Relative regional vulnerability of water resources to climate change (JAWRA, v35, No. 

6). 

Haak A., J.E. Williams, H.M. Neville, D.C. Dauwalter, and W.T. Colyer. 2010. Conserving peripheral 

trout populations: the values and risks of life on the edge. Fisheries Management 35(11): 530-548. 

Isaak, D.J., C.H. Luce, B.E. Rieman, D.E. Nagel, E.E. Peterson, D.L. Horan, S. Parkes, and G.L. 

Chandler. 2010. Effects of climate change and wildfire on stream temperatures and salmonid thermal 

habitat in a mountain river network. Ecological Applications 20(5): 1350-1371. 

Isaak, D.J., B.E. Rieman, and D.L. Horan. 2008. A bull trout monitoring protocol and example 

application in the Secesh River Basin. Rocky Mountain Research Station publication. 

www.fs.fed.us/rm/boise/AWAE/projects/stream_temperature.shtml 

Knowles, N, Dettinger, M., and D. Cayan. 2006. Trends in snowfall versus rainfall in the western US 

(J. of Climate). 

Mote, P.W. 2003. Trends in temperature and precipitation in the PNW during the 20 th century (NW 

Science v77, No4). 

Neville, H., J. Dunham, A. Rosenberger, J. Umek, B. Nelson. 2009. Influences of Wildfire, Habitat 

Size, and Connectivity on Trout in Headwater Streams Revealed by Patterns of Genetic Diversity. 

Transactions of the American Fisheries Society 138: 1314-1327. 

Rieman, B.E., D. Isaak, S. Adams, D. Horan, D. Nagel, and C. Luce. 2007. Anticipated Climate 

Warming Effects on Bull Trout Habitats and Populations Across the Interior Columbia River Basin. 

Trans. American Fisheries Society 136(6): 

Rieman, B.E., J.D. McIntyre. 1993. Demographic and Habitat Requirements for Conservation of bull 

trout. USDA Intermountain research General Technical Report INT-302. 



Rieman, B.E., D.J. Isaak, S. Adams, D. Horan, D. Nagel, C. Luce, D. Myers. 2007. Anticipated 

climate warming effects on bull trout habitats and populations across the Interior Columbia River Basin. 

Transactions of the American Fisheries Society 136: 1552-1565. 

Spruell, P., F.W. Allendorf. 1997. Nuclear DNA analysis of Oregon bull trout. Final report for ODFW 

Report 97/5. 

Wenger, S.J., Luce, C.H., Hamlet, A.F., Isaak, D.J., and H.M. Neville. 2010. Macroscale hydrologic 

modeling of ecologically relevant flow metrics. Water Resource Res., Vol. 46, WO19513. 10 p. 




Ouachita National Forest 

March, 2012 

Prepared by: 

J. Alan Clingenpeel 

Forest Hydrologist 


Hot Springs, Arkansas 


Ouachita National Forest Watershed Vulnerability Assessment, Southern Region (R8) 

FOREST CONTEXT 

The Ouachita National Forest covers over 1.7 million acres in western Arkansas and eastern Oklahoma, 

and is located within the Southern Region (R8) of the USFS. The forest is primarily composed of 

shortleaf pine and hardwoods and is largely within the Ouachita Mountain Ecoregion with some 

ownership in the Arkansas Valley and Mid Coastal Plains - Western Ecoregion. The Ouachita Mountains 

form the backbone of the forest with an east-west orientation. Weather patterns for the Ouachita 

Mountains in Arkansas and Oklahoma are characterized by a temperate climate due to its location in the 

center of the North American continent. Air masses that move across the national forest generally 

originate from the Eastern Pacific Ocean, Western United States, the Gulf of Mexico, and Canada. The 

sources of moisture for the region are the Pacific Ocean and the Gulf of Mexico. Because of the general 

circulation characteristics of the atmosphere, weather systems generally move from west to east across the 

Ouachita Mountains (USDA Forest Service, 1999). Mean annual precipitation ranges from 39.4 inches 

per year (Fort Smith, AR) in the northwestern area of the forest to 55.5 inches per year (Hot Springs, AR) 

in the southeastern areas of the forest. Corresponding surface runoff values range from 14 to 22 inches per 

year. 

PARTNERS 

The forest was fortunate in that a subwatershed 

analysis was recently completed with the Travel 

Management Project. In addition, the climate 

change study included consultations with Bill 

Elliot (Rocky Mountain Research Station), Dan 

Marion (Southern Research Station), and Steve 

McNulty (Southern Research Station). Data for 

climate scenarios was taken from the TNC 

Climate Wizard website 

(http://www.climatewizard.org/). 



ASSESSMENT OBJECTIVE 

The assessment objective, using the Aquatic Cumulative Effects (ACE) model, is to determine changes in 

risk level for aquatic biota for each subwatershed for two climate scenarios (B1 and A1B) for the near 

term (year 2050) and long term (year 2080). 


There are 13 fourth-level 

cataloging units on the forest 

and 50 fifth-level watersheds. 

Within those fifth-level units, 

190 sixth-level subwatersheds 

have some NFS ownership. The 

area assessed included all NFS 

ownerships under the 

management of the Ouachita 

National Forests. Subwatersheds 

are also referred to as sixth-level 

watersheds or 12 digit 

hydrologic units. They are 

typically 10,000 to 40,000 acres 

in size. 

CONNECTIONS TO OTHER ASSESSMENTS, PLANS AND EFFORTS 

This analysis has several connections within the Forest and across the Region. The Forest has participated 

in a number of assessments at various scales. The first and largest assessment was the Ozark Ouachita 

Highlands Assessment (OOHA) Aquatic Condition report (USDA Forest Service, 1999). This assessment 

addressed water quality and management concerns across a three state area at the fourth-level cataloging 

units (eight digit hydrologic units). 

From 1999 through 2001, the Region (including the Ouachita) completed a series of forest-level 

assessments using the East-wide Watershed Assessment Protocol (EWAP, 2000). This assessment 

occurred at the fifth-level watershed scale. It addressed a number of conditions and vulnerabilities for 

each watershed and applied a ranking system for condition, vulnerability, and overall watershed health 

among the fifth level watersheds on the forest. 

From that exercise, watershed condition was determined for many forest-level plan revisions across the 

region. The Ouachita was one of the forests that took the information from the assessments and developed 

a disturbance (based on sediment) model to address cumulative effects. The value of the model was that it 

provided a correlation of disturbance to fish guild communities. For the first time, this allows a numerical 

assessment of the effect of management actions on fish communities. Again this exercise was at the fifthlevel 

watershed. To date, this process has been applied on 10 of 16 forests in the Southern Region. 

The Ouachita NF developed a project level analysis using the same protocols found in the forest plan. 

This model is referred to as the Aquatic Cumulative Effects (ACE) model. This forest level model was 

modified to address the short and long term risks of climate change for two different climate scenarios. 



WATER RESOURCES 

The ACE model is a disturbance model that uses changes in sediment to compare various management 

scenarios and determine the effect on aquatic biota. Model inputs include the following. 

• Watershed layer 

• Current land use (grid) 

• Ecoregion (section level) 

• Ownership (forest service or other) 

• Slope class (derived from dems) 

• Roads and trails (ownership, maintenance level and surface) 

• Recreation use (motorized recreation use) 

For terrestrial sediment yields; land use, ecoregion, slope class, and recreation use were summarized by 

30 meter grids. An erosion coefficient (pounds/acre/year) was determined for each grid combination and 

the grids were accumulated for each subwatershed. Sediment was determined using Roehl (1962). Roads 

and trails were identified by ownership, maintenance level, ecoregion, and recreation use level. A 

sediment coefficient (tons/mile/year) was determined from Water Erosion Prediction Project (WEPP, 

1999) surveys for each road or trail combination. The roads and trails were clipped by subwatershed and 

summarized by total miles of each combination. 

EXPOSURE (CLIMATIC CHANGES) 

Predictive Models Used 

The forest ACE model was used to establish current condition and potential current condition (assuming 

fully funded and implemented road and trail maintenance). The ACE model calculates general land uses 

and linear events (roads and trails) separately. 

From the TNC climate wizard, changes in precipitation and temperature were captured by month from the 

composite climate change models. The changes in climate were used to modify the climate generator in 

WEPP. Roads and trails coefficients were reanalyzed in WEPP Road to determine changes in sediment 

production from roads and road use levels. Because of the time consuming nature of recalculating 

individual climates, a proportional relationship for road and trail sediment increases was used. 

The Universal Soil Loss Equation (USLE) (Dissmeyer and Foster, 1984) was used for terrestrial 

coefficients. The R factor was modified using information from Phillips (1993). The new R value for the 

climate change scenarios was used in the USLE equation. Results were proportionally distributed for 

terrestrial coefficients. 

Storm intensity was determined for roads and trail by reducing the number of days of precipitation in the 

climate generator model. In theory, this should force the generator to predict more intense storms. The 

value used was half of the percent change in precipitation volumes (personal communication, Bill Elliot). 

Anticipated Climate Change 



The table below shows the monthly and annual changes predicted for the B1 and A1B climate scenarios. 

This is an average of all of the climate generated models (CGM). The Forest should experience a 2 to 4 

degree F increase in the B1scenario in 2050 with an additional 1 to 2 degree F increase to 2080. The 

largest temperature increase will occur in the summer months and early fall. The 2050 A1B shows a 4 to 

5 degree F increase throughout the year with an additional 2 degree F increase by 2080. 

Precipitation values are mixed with increases and decreases. Monthly declines are anticipated for all 

months except April, August, and December for the 2050 B1 scenario. Annually, a two percent reduction 

is anticipated for both near term (2050) and long term (2080). The 2050 A1B scenario is similar with a 

three to four percent reduction with the greatest reduction in precipitation occurring in summer and late 

fall. Storms are forecast to be more intense for both scenarios. However, that value was not quantified. 

Increases in Temperature (°F) Percent change in precipitation (inches) 

B1 2050 B1 2080 A1B 2050 A1B 2080 

January 2.70 4.42 4.38 6.00 

February 3.50 4.01 4.46 5.19 

March 3.46 4.25 4.70 5.74 

April 2.99 4.46 4.49 5.93 

May 3.68 4.48 5.02 7.16 

June 3.90 4.64 5.34 7.04 

July 4.14 4.98 5.40 7.28 

August 4.13 5.04 5.21 6.84 

September 4.23 5.49 5.35 7.45 

October 4.12 5.46 5.29 7.15 

November 3.52 4.36 4.93 6.15 

December 3.18 4.40 4.11 5.97 

Annual 3.63 4.67 4.89 6.49 


B1 2050 B1 2080 A1B 2050 A1B 2080 

(0.69) 8.85 5.98 1.68 

(0.97) (4.50) (2.54) (1.24) 

(0.75) (4.30) 0.63 (5.17) 

5.42 2.45 (1.19) 0.67 

(8.46) (1.28) (6.26) (10.68) 

(5.87) (7.17) (8.76) (12.37) 

(8.34) (2.70) (7.39) (12.84) 

1.20 6.97 1.52 2.61 

(0.49) 1.10 (3.47) 1.32 

(13.81) (8.17) (9.75) (8.17) 

0.91 (5.08) (7.93) (8.75) 

5.20 (9.39) (1.69) (1.68) 

(2.22) (1.93) (3.40) (4.55) 

Changes to key hydrologic processes and their direct and secondary impacts to each water resource 

Using the new climates from TNC climate wizard and batch runs from WEPP, a 7 to 13 percent increase 

in sediment from linear disturbances (roads and trails) were identified for the various road types, climate 

scenarios, and time periods. 

From the modified R values, a 3 percent increase in average annual erosion for the B1 scenario (both year 

classes) and 15 percent increase in average annual erosion for the A1B (both year classes) was used. This 

data is somewhat suspect because of the scale used for the R values, the limited number of CGMs, and the 

improvements in climate predictions since the early 1990s. 

WATERSHED RISK 

Stressors that amplify the anticipated hydrologic changes 

Many of the stressors are natural or historic; geology, erodible soils, steepness, and vegetation types are 

all natural features of a watershed that may or may not amplify hydrologic changes. Past human activities 

may also have a bearing. Certainly stressors that have been chosen to describe climate change (increases 

in temperature and storm intensity as well as fluctuations in precipitation) are stressors. However human


activities, such as past and current land use and roads and trails, are factors directly affecting hydrologic 

change within a watershed. 

Buffers that modify the anticipated hydrologic changes 

Land use and changes in land use is a useful tool to anticipate changes in sediment. This is the primary 

vehicle used in the ACE model to address cumulative effects. For the purposes of this exercise, the 

current land use condition was frozen for both scenarios and time frames. In addition, forest management 

was not addressed. No forest management activities (e.g. clearcuts or thinnings) were modeled. 

Roads and trails (including their current condition and level of use) is the other useful stressor to address 

changes in sediment yield. Currently, many forest roads on the Ouachita National Forest are seeing 

increased off highway vehicle (OHV) use and substantial reductions in maintenance. Bringing these forest 

roads/trails up to an acceptable level of construction standard and providing maintenance is the easiest 

way to buffer sediment losses. Reducing user created trails is another method to buffer sediment losses. 

For this exercise, the current road and trail condition and potential current condition (assuming roads and 

trails built to standard and maintained) were used in the climate change predictions. 

Other methods not addressed could include 

reducing road and trail miles (obliteration 

or maintenance level 1) or reducing the 

numbers of OHV users. County road 

maintenance and design could also be 

addressed and improved. 

Method used to characterize watershed 

risk 

Increases in sediment can directly affect 

stream habitats by reducing available 

substrate, and reducing pool volumes and 

pool depths. Indirectly changes 

in habitat can affect fish 

communities. The sensitivity of 

these changes was established by 

taking known fish population 

samples and determining the 

annual sediment contribution 

from the watershed above the 

sample location. Percent 

sediment increase (over a 

baseline condition) was 

compared to the relative 

abundance of various fish guilds. 

Ecoregions and slope (how steep 

the watershed is) were used to 

generate broad categories. When 

a wedge pattern was found, 

sensitivity thresholds were 



identified through quadrants. These quadrants were then used to evaluate watershed health and the 

potential risk to fisheries from increases in sediment (green is a low risk, yellow is a moderate risk, and 

red is a high risk. 

RESULTS 

The following map shows the surface ownership 

for the Ouachita National Forest and the sixthlevel 

subwatersheds associated with that 

ownership. 

The map below identifies the subwatershed risk 

levels for the current condition and the potential 

Two factors exist for this analysis. 

The first is that the Forest has not 

implemented its Travel Analysis. 

This means that the forest floor is 

still open and that user created 

trails still exist. The second factor 

is that the maintenance level 1 and 

2 roads and motorized trails are not 

being maintained and have fallen 

below an acceptable road 

construction standard. 

To demonstrate the ability of the 

model to respond to change, the 

model was recalibrated to assume 

that the roads and trail systems 

were brought up to the forest 

standard for construction and 

maintenance. The map to the right 

shows the difference between the 


risk to aquatic biota. This analysis was 

taken from the Travel Management 

Assessment that the forest completed in 

January of 2010. Green subwatersheds are 

low risk to aquatic biota, yellow are 

moderate, and red are high risk. This 

assessment found 88 subwatersheds with a 

high risk, 46 with a moderate risk and 56 

with a low risk.


current condition and a condition with road maintenance and the forest floor closed to OHV use. All 

subwatersheds show improvement. Some subwatersheds show enough improvement to move to a lower 

risk category. The dark green subwatershed would actually move from a high risk to a low risk and five 

other subwatersheds would move from a high risk to a moderate risk. Eleven subwatersheds would move 

from moderate to low risk (light green). 

Current Condition and B1 

The B1 scenario for 2050 found that an 

additional four subwatersheds would move 

from a moderate risk to a high risk (shown in 

dark red) and that one subwatershed would 

move from a low risk to a moderate risk 

(shown in red). Comparing the current 

condition for 2080 B1 scenario provided the 

same results. There was no change for the B1 

scenario between the near term and long 

term predictions. 

Current Condition and A1B 

The current condition and A1B predicts a 

poorer condition than B1. There are 16 

subwatersheds that moved from a moderate 

risk to a high risk for aquatic biota. In 

addition, 15 subwatersheds moved from a 

low risk to a moderate risk. The long term 

climate change prediction (2080) is worse 

with an additional subwatershed moving 

from a low risk to a moderate risk. 

CONCLUSIONS 

The predicted climate changes from TNC 

climate wizard and their application to 

WEPP is a useful tool to predict different 

climate scenarios. The use of Phillips (1993) 

was not as useful because of the scale the 

data is represented at and improvements in 

climate predictions from the early 1990s. 

The current Forest watershed condition has 88 watersheds with a high risk and 46 with a moderate risk. 

The simple act of maintaining of roads, bringing them up to plan standards, and limiting recreation use 

can reduce the number of subwatersheds with high risk by six. The number of subwatershed with a 

moderate risk would decrease by 11. Seventeen subwatersheds (almost 10 percent of all subwatersheds) 

would move from a higher risk category to a lower risk category by complying with the forest plan (road 

and trail standards) and providing maintenance. Over time, all of the various scenarios suggest an 

increased risk to aquatic biota. There are many approaches to managing that risk, the least of which is to 

provide maintenance. 



Scenario 

2010 

Current 

2010 

Mngt 

resp* 

2040 

B1 

2040 

B1 

Mngt 

resp 

2080 

B1 

2080 

B1 

Mngt 

resp 


2040 

A1B 

2040 

A1B 

Mngt 

resp 

2080 

A1B 

Risk 

High 88 82 93 85 93 85 105 96 105 96 

Moderate 46 40 42 43 42 43 44 43 45 43 

Low 56 68 55 62 55 62 41 51 40 51 

*Mngt resp – responsible management that brings roads and trail up to FS standards 

APPLICATION 

2080 

A1B 

Mngt 

resp 

This project is applicable at the sixth-level subwatershed scale. Conceivably, it is applicable at the fourth 

and fifth level scales as well. However, the risk levels would have to be reevaluated at the fourth-level 

basin scale. 

The information exists for application across the south – many forests have established aquatic thresholds 

by ecoregion. It is also applicable on the project level when used at the subwatershed scale. 

CRITIQUE 


• This approach uses thresholds for fish. Other aquatic biota such as mussels are more sensitive to 

changes in sediment. 

• This particular exercise did not include water yield and regimen which could easily provide 

additional stress to aquatic biota. 

• The analysis is based on averages. Extreme events such as droughts or floods which would 

modify aquatic and riparian habitats were not taken into account. 


• TNC climate wizard 

− user friendly 

− multiple scenarios with multiple GCMs 

• WEPP climate generator 

− Individual sites are easily modified 

− A national application for the lower 48 states 


• The USLE R-factor. Given more time or knowledge, I would have recalculated those values. 

This was the weakest part of the analysis. 


• TNC climate wizard 

• WEPP climate generator 

• ArcView and ArcMap



Alan Clingenpeel 


(501) 321-5246 

aclingenpeel@fs.fed.us 

REFERENCES 

Dissmeyer, G.E., and G.R. Foster. 1984. A guide for predicting sheet and rill erosion on forestland. 

USDA For. Serv. Gen. Tech. Publ. R8-TP 6. 40. 

EWAP. 2000. http://www.fs.usda.gov/Internet/FSE_DOCUMENTS/stelprdb5291793.pdf. 

Phillips, D.L., D. White, and C.B. Johnson. 1993. Implications of climate change scenarios for soil 

erosion potential in the United States. Land Degradation and Rehabilitation 4: 61-72. 

Roehl, J. W. 1962. Sediment source areas, delivery ratios, and influencing morphological factors. IASH 

Comm of Land Eros, Pub 59:202-213. 

USDA Forest Service. 1999. Ozark-Ouachita Highlands Assessment: aquatic conditions. Report 3 of 5. 

GTR SRS-33. Asheville, NC: USDA, Forest Service, Southern Research Station. 317 p. 

WEPP. 1999. http://forest.moscowfsl.wsu.edu/fswepp/docs/fsweppdoc.html 




Chequamegon-Nicolet National Forest 

July, 2012 

Prepared By: 

Dale Higgins 

Hydrologist 


Park Falls, WI 


Chequamegon-‐Nicolet National Forest Watershed Vulnerability Assessment, Eastern Region (R9) 

INTRODUCTION 

Maintaining and restoring watershed resilience is an appropriate strategy for responding to climate change 

because changes are anticipated to affect every component of the hydrologic cycle. But watersheds can 

differ greatly in their vulnerability to climate change. Understanding differences in watershed 

vulnerabilities is necessary to develop adaptive management strategies and implement targeted land 

management practices. 

Several National Forests, representing each region of the US Forest Service, are working to assess the 

potential impacts of climate-induced hydrologic change on important water resources. Each forest is 

identifying important water resources, assessing their exposure to climate change, evaluating risk, 

categorizing watershed vulnerability, and recommending potential management responses. 

The Chequamegon-Nicolet National Forest (CNNF) is one of the pilot Forests. This report summarizes an 

assessment of watershed vulnerability associated with four important water resources: wetlands, 

groundwater recharge, stream fishes and infrastructure (culverts at road stream crossings). More detailed 

individual reports are available for each of these resource assessments. 

These four resources were selected because of their importance to people and the local environment. 

Wetlands (with an emphasis on bogs) were selected because of their importance to the northern 

Wisconsin landscape and their apparent vulnerability to increased potential evapotranspiration. 

Groundwater recharge was selected because of the importance of groundwater to the ecology of many 

streams, lakes, and wetlands; the potential for changes associated with higher evapotranspiration; and to 

take advantage of a groundwater inventory currently underway on the Forest. The ultimate goal will be to 

model the projected effects of changes in groundwater recharge on aquifer levels, flow paths and flow 

rates and to evaluate those effects on surface water resources. Wetlands and groundwater recharge were 

also selected because they were unlikely to be addressed by the other National Forests in the pilot. 

Infrastructure was selected because there is a concern that precipitation frequency and intensity may 

increase in the future, threatening culverts that are not properly sized. This is one of the most urgent 

management considerations because culverts installed now need to last up to 100 years. Stream fish-water 

temperature was selected because of the potential for future stream temperature increases and the 

subsequent effects on cold and cool water fish. It was also selected because there was an opportunity to 

take advantage of a statewide analysis of the potential effects of climate change on stream fish in 

Wisconsin. 

METHODS 

Methods are summarized here; more detail is provided in the following sections. In all cases, the 

assessment included two basic steps: (1) some type of modeling to characterize the potential effect or risk 

of projected climate change on the water resource, and (2) extrapolation of that potential risk to 

characterize the vulnerability of that resource at the watershed scale. The five individual vulnerability 

ratings (wetlands, groundwater recharge, infrastructure, cold water fish, and cool water fish) were 

combined into one composite numerical watershed vulnerability ranking with the following thresholds: 

1.0, very low; 1.2-2.4, low; 2.6-3.0, moderate; and 3.2-4.0, high. The composite rankings were based on 

averages of the individual resource ratings. 

Climate data required for modeling were obtained from the Wisconsin Initiative on Climate Change 

Impacts (WICCI) program (www.wicci.wisc.edu/). The WICCI Climate Working Group has developed a 

regional-scale, daily dataset of historical and future projections of total precipitation, and maximum and 

minimum temperature for the time period 1950-2099 at an 8-km spatial resolution across Wisconsin. This 



data is available for 14 global circulation models (GCMs) and three future scenarios for greenhouse gas 

emissions (A2, A1B, B1). It was developed by downscaling the coarse-scale climate projections of the 

GCMs. The ideal approach for climate change analyses would be to model the effects for all 14 GCMs 

and all three scenarios to evaluate the full range of potential climate change impacts. Given limited time 

and resources, this assessment used just one GCM, the GFDL-CM2.0, and one scenario for one pixel of 

data located on the Park Falls unit of the CNNF. The A1B scenario was selected because it provides an 

intermediate level of greenhouse gas emissions relative to the other scenarios. 

Wetlands 

Potential changes in wetland hydrology were determined using the Peatland Hydrologic Impact Model 

(PHIM) (Guertin et al. 1987; Brooks et al. 1995). PHIM is a physically-based, continuous simulation 

model for predicting water yield and streamflow from peatland and upland watersheds typical of the 

northern Great Lakes region. 

The PHIM was run with 40 years of historic climate data (1961-2000) and 20 years of projected climate 

data (2046-2065). The potential effect of climate change on bog hydrology was evaluated by determining 

differences in average annual and seasonal runoff and evaporation from the upland-peatland complex, and 

average annual and seasonal water level in the bog. The results were extrapolated to all HUC-6 

watersheds encompassing the National Forest based on the proportion of total wetland and acid wetland 

in each HUC-6 watershed. 

Groundwater Recharge 

The groundwater recharge portion of the analysis focused on the Park Falls unit of the Forest to take 

advantage of a recently initiated project characterizing groundwater resources on this portion of the 

Forest. This project is being conducted by the Wisconsin Geological and Natural History Survey 

(WGNHS) and United States Geological Survey (USGS). 

Potential changes in groundwater recharge were determined for the Park Falls unit using the Soil Water 

Balance Model (SWBM) (Westenbroek et al. 2010; Dripps and Bradbury 2007). The SWBM estimates 

recharge using gridded watershed data and tabular climatic data. The watershed data include soil water 

capacity, hydrologic soil group (HSG), flow direction, and land use. 

The results of the Park Falls modeling were extrapolated to all HUC-6s encompassing the National Forest 

based on the proportion of HSG in each HUC-6. Watersheds with no or reduced recharge were considered 

most vulnerable while those with increases in recharge were considered least vulnerable or most resilient. 

Infrastructure-Culverts 

The analysis included four primary steps: (1) evaluating climate change projections to determine the 

potential for increases in flood magnitudes, (2) reviewing culvert sizing criteria and hydraulic modeling 

results, (3) determining road-stream crossing density and runoff potential for HUC-6s within the CNNF, 

and (4) characterizing the vulnerability of HUC-6s to increased flood flows and failure of culvert 

infrastructure based on steps 1-3. 

WICCI summary data were evaluated for evidence that flood flows may increase in the future. Key data 

used for this evaluation were projections for the frequency of 1-, 2-, and 3-inch rainstorms and for annual 

and seasonal precipitation and air temperatures. 



Culvert sizing criteria were obtained from the CNNF Forest Plan (USDA Forest Service 2004) and 

Stream Simulation (USDA Forest Service 2008) guidelines. The results of hydraulic modeling for a select 

number of recent culvert replacements on the CNNF were reviewed and compared to the culvert sizing 

criteria. These included several sites with low to moderate runoff potential and one with high runoff 

potential. 

The number of road-stream crossings and their density (#/sq mi) within the CNNF boundary were 

determined from an inventory conducted by the CNNF. The watersheds were placed into one of four 

classes based on road-stream crossing density. Runoff potential was estimated from hydrologic soil 

groups. Watersheds were placed into one of four classes based on their average HSG rating. 

The vulnerability of individual HUC-6s to increased flood flows and failure of culvert infrastructure was 

estimated by combining the road-stream crossing density and runoff potential classes. The ratings for 

these two parameters were combined to classify the vulnerability of each HUC-6 as either very low, low, 

moderate, or high. In this classification, HSG ratings were given twice the weight of crossing density 

ratings because HUC-6s with high runoff potential were expected to experience higher increases in flow, 

making infrastructure in those watersheds more vulnerable than watersheds with low runoff potential, 

regardless of the crossing density. 

Stream Fishes 

The analysis included two primary steps: (1) evaluating statewide modeling of the potential impacts of 

climate warming on stream fish distributions at the Forest level, and (2) summarizing those results to 

characterize the vulnerability of cold and cool-transitional stream fishes to climate change at the 

watershed scale. 

Lyons et al. (2010) analyzed the potential effects of climate change on water temperature and 50 stream 

fishes in Wisconsin. They utilized habitat models developed from the Wisconsin aquatic gap program to 

estimate existing and future distributions of each fish. These models were applied to 86,898 km of stream 

(at the 1:100,000 scale) in Wisconsin under four different climate scenarios, including current conditions, 

minor warming (summer air temperature increases 1 °C and water 0.8 o C), moderate warming (air 3 o C 

and water 2.4 o C) and major warming (air 5 o C and water 4.0 o C). The water temperature increase of 

0.8 o C for each 1.0 o C increase in air temperature used in their study was an oversimplification 

necessitated by the statewide study that did not take into account how groundwater input, land uses, or 

changes in flow might alter the response of streams to air temperature increases. 

For the CNNF analysis, the GIS layers of predicted fish distributions developed by Lyons et al. (2010) 

were obtained for 15 fish species from the Wisconsin Department of Natural Resources (WDNR) and 

USGS. The selected species included 2 cold water fishes (brook trout and mottled sculpin), 8 cool or 

transitional water fishes (blacknose dace, brook stickleback, creek chub, longnose dace, northern 

hogsucker, northern redbelly dace, walleye, white sucker) and 5 warm water fishes (black crappie, 

hornyhead chub, logperch, smallmouth bass, and stonecat). The distributions for each climate scenario 

and species were intersected with CNNF HUC-6 delineations. The amount of predicted habitat for the 

current climate and moderate warming was determined for each species by HUC-6 and for all HUC-6s 

combined. One additional cold water species, brown trout, was modeled but not carried through the 

analysis. 

The vulnerability of individual HUC-6s was estimated by determining the percentage change in habitat 

for each species in the watershed. That percentage was based on the total habitat for all HUC-6s for that 

species. Within each HUC-6, cold and cool water species were combined by calculating a simple 

arithmetic average. Each HUC-6 was then classified according to its vulnerability to climate change 



impacts to cold and cool water species by developing and applying thresholds for average change in fish 

distribution. 

EXPOSURE 

Northern Wisconsin has a typical continental climate with cold winters and warm summers. Precipitation 

averages 32 inches per year, two-thirds of which falls during the growing season. Snowfall generally 

averages 50 to 60 inches per year but some localized areas receive 70 to 140 inches. There are normally 

110 to 130 days with snow cover greater than 1 inch. Evapotranspiration and runoff average 20 inches 

and 12 inches per year, respectively. Average annual temperature is 40 o F (4.4 o C) with a January average 

of 10 o F (-12.2 o C) and July average of 66 o F (18.9 o C). 

The WICCI downscaled data from 14 GCMs for the A1B scenario projects that northern Wisconsin will 

likely experience an increase in average annual air temperature of 6.5 o F (3.6 o C) by the mid-21 st century 

(Figure 1). Warming will be most pronounced in winter (increase of 8.5 o F, 4.7 o C) and least pronounced 

in summer (increase of 6.5 o F, 3.6 o C) (Figure 2). Average annual precipitation is expected to increase by 

2.0 inches with most of the increase occurring in fall, winter, and spring (Figure 3). 

Figure 1. Projected increase in average annual air temperature for WI, A1B scenario 



Figure 2. Projected increase in seasonal air temperatures for WI, A1B scenario 

Rainfall intensity is expected to increase. The number of days with precipitation greater than 2 inches is 

expected to increase from seven days per decade to about 9.5 or 10 days per decade (Figure 4). Much of 

this increase is projected to occur in spring and fall (Figure 5). The frequency of storms producing more 

than 3.0 inches of rainfall in 24 hours is also expected to increase, especially in spring and fall. There will 

also be a shorter snow season with less snowfall and snow depth. 

The GFDL-CM2.0 model produced average annual temperatures for the historic and future periods of 4.6 

0 C (40.3 o F) and 8.1 o C (46.6 o F), respectively (Table 1). Average annual precipitation was predicted to 

increase by 0.8 inches or 2.6 percent from 31.1 to 31.9 inches (Table 1). Average monthly precipitation 

would increase by about 0.5-1.5 inches in January, March, April, and May and decrease a similar amount 

in June, July, and October (Figure 6). 



Figure 3. Projected change in average annual precipitation for WI, A1B scenario 

Figure 4. Projected increase in days with 2” precipitation events in WI, A1B scenario 



Figure 5. Projected increase in 2”-24” precipitation by month for WI 

Average Air Temperature ( o C) 

25 

20 

15 

10 

5 

0 

-‐5 

-‐10 

-‐15 

1960-‐2000 

2046-‐2065 

Peatland Hydrologic Impact Model (PHIM) 

Calibrated S2 Bog WS from Marcel Exp Forest in MN 

WICCI Climate Data for longitude -‐90.1, ladtude 45.8 

(located on Park Falls unit of Chequamegon-‐Nicolet NF) 

GFDL_CM2.0 Model, Scenario A1B 

Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec 

Month 

Figure 6. Average monthly precipitation for PHIM runs for Park Falls unit 



RESULTS 

Wetlands 

For the historic period, the PHIM produced an average monthly snowpack that peaks in March at 2.2 

inches of water equivalent and normally melts by mid-April (Figure 7). With warmer winters in the 

future, PHIM projects that average monthly snow water equivalent would peak in February at 1.7 inches 

and melt by mid-March. This represents a decline in average snow water of nearly 25 percent with melt 

occurring about one month earlier. 

The modeling results indicate average annual evapotranspiration from the upland-peatland complex 

would increase by 3.2 inches (from 21.7 to 24.9 inches), a 15 percent increase (Table 1). Average annual 

runoff would decline by 1.3 inches (from 5.5 to 4.2 inches), which represents a 24 percent decline. From a 

seasonal standpoint, runoff would remain the same in winter, increase in spring by 0.4 inches, and 

substantially decline in summer and fall (Table 1). 

Average annual water levels would decline only slightly in the bog but changes for individual seasons and 

months would be much greater. Average annual water levels in the bog would decline from 9.5 to 8.1 

inches, or about 15 percent (Table 1). Monthly water levels would be unchanged in Jan-Feb, increase 0.5- 

1.25 inches in Mar-May, and decline 0.5-4.5 inches in Jun-Dec (Figure 8). No flow days were predicted 

to occur 4.4 percent of the time (16 days/yr) for the current climate but would increase to 23.4 percent of 

time (85 days/yr) under the climate change scenario. The 4.5-inch decline in water levels in August and 

September and large increase in no-flow days could have a substantial effect on plant communities and 

carbon processes in the bog. 

Ave Snow Water Equivalnet (inches) 

2.5 

2.0 

1.5 

1.0 

0.5 

0.0 


Month 

1960-‐2000 

2046-‐2065 

Figure 7. PHIM average monthly watershed snow water equivalent for 1961-2000 and 2046-2065 

The results indicating earlier snowmelt and higher initial water levels in the spring are similar to the 

results obtained by McAdams et al. (1993) who used PHIM to model streamflow and water table changes 

in the S2 bog due to climate change. S2 is an experimental peatland watershed located on the Marcell 

Experimental Watershed in northern Minnesota. The researchers used temperature and precipitation 


Peatland Hydrologic Impact Model 

Calibrated S2 Bog WS 

from Marcel Exp Forest in MN 

WICCI Climate Data 

longitude -‐90.1, ladtude 45.8 

(located on Park Falls unit 

of Chequamegon-‐Nicolet NF) 

GFDL_CM2.0 Model, Scenario A1B


increases projected for northern Minnesota by the GISS global climate model at the time of their study. 

These included monthly increases of 3 to 6 o C for temperature and 5 to 25 percent for precipitation. In 

their case, though, growing-season water levels in the bog were projected to decline by only 0.2-0.6 

inches because higher evapotranspirational losses would be offset by higher summer precipitation. 

There was one modeling problem that remained unresolved. The spring runoff hydrograph for the historic 

period appears to peak at about 25 to 50 percent of expected runoff during the spring snowmelt season 

(Figure 9). It also appears to produce slightly higher runoff than expected in the fall. 

While this modeling problem causes some concern, the overall results seem to provide reasonable 

estimates of the potential impacts of climate change on bog hydrology in northern Wisconsin. These 

include future increases in average annual evapotranspiration of about 3.2 inches, decreases in runoff of 

1.3 inches (about 25 percent) with an increase in spring and decreases in summer and fall, and lower 

water levels in the bog in summer and fall of 2-4.5 inches with an increase in no-flow days. 

Although the ecological implications of these potential changes in wetland hydrology need further 

evaluation, for the purposes of this analysis they were considered sufficient to conclude that climate 

change poses some risk to the Forest’s wetlands in general and to bogs in particular. These risks include 

loss of wetland area, changes in wetland plant communities, and alteration of wetland processes such as 

water chemistry, peat accumulation, and geochemical cycling. 

Season 

Time 

Period 

Air 

Temp. 

( o C) 

Ppt. 

(in) 

ET 

(in) 


RO 

(in) 

Water 

Level 

(in) 

Winter 1961-2000 -10.3 3.2 0.2 0.4 8.6 

2046-2065 -7.0 3.9 0.4 0.4 8.2 

Spring 1961-2000 4.3 7.5 4.7 1.6 9.5 

2046-2065 7.6 9.7 6.0 2.0 10.3 

Summer 1961-2000 17.9 11.7 12.5 1.3 9.3 

2046-2065 22.3 10.3 14.8 0.8 6.7 

Autumn 1961-2000 6.3 8.7 3.7 2.1 10.5 

2046-2065 9.2 8.0 4.4 0.8 7.3 

Annual 1961-2000 4.6 31.1 21.7 5.5 9.5 

2046-2065 8.1 31.9 24.9 4.2 8.1 

Table 1. Average seasonal and annual water balance components from modeling of potential climate change 

impacts to wetlands on the Chequamegon-Nicolet National Forest. WICC climate data for longitude 90.1, latitude 

45.8 located on Park Falls Unit of Chequamegon-Nicolet NF, GFDL_CM2.0 Model, A1B scenario. Water level 

estimates from Peatland Hydrologic Model (PHIM).


Water Level (inches) 

Figure 8. PHIM monthly average bog levels for 1961-2000 and 2046-2065 

Classification of watershed vulnerability to wetland impacts from climate change was based on the 

proportions of total wetland and acid wetland within the National Forest boundary of each HUC-6 (Figure 

10). Three risk categories were established for both total and acid wetlands. The percentage of total 

wetland area ranged from 0 to 55.8 percent. Those with less than 10 percent were rated low, 10 to 30 

percent were rated moderate, and greater than 30 percent were rated high. Acid wetland ranged from 0 to 

42.8 percent of the area for all HUC-6s. Those with less than 5 percent were rated low, 5 to 15 percent 

were rated moderate, and greater than 15 percent were rated high. These two risk classes were combined 

to form one vulnerability classification for each watershed, as indicated in Table 2. 

Runoff (inches) 

12.0 

11.0 

10.0 

9.0 

8.0 

7.0 

6.0 

5.0 

4.0 

1.0 

0.9 

0.8 

0.7 

0.6 

0.5 

0.4 

0.3 

0.2 

0.1 

0.0 

Peatland Hydrologic Impact Model 

Calibrated S2 Bog WS 

from Marcel Exp Forest in MN 

WICCI Climate Data 

longitude -‐90.1, ladtude 45.8 

(located on Park Falls unit 

of Chequamegon-‐Nicolet NF) 



1960-‐2000 

2046-‐2065 

Month 

Figure 9. PHIM average monthly bog runoff for 1961-‐2000 and 2046-‐2065 


1960-‐2000 

2046-‐2065 

Peatland Hydrologic Impact Model (PHIM) 

Calibrated S2 Bog WS from Marcel Exp Forest in MN 

WICCI Climate Data for longitude -‐90.1, ladtude 45.8 

(located on Park Falls unit of Chequamegon-‐Nicolet NF) 



Month


The relative vulnerability of each HUC-6 to climate impacts on wetlands is presented in Figure 11. There 

were 38 watersheds with low vulnerability because of low percentages of both total and acid wetlands. 

There were 82 HUC-6s classified as having moderate vulnerability. There were 19 watersheds classified 

as having high vulnerability and also 19 watersheds classified as having very high vulnerability because 

of high percentages of both total and acid wetlands. They are located primarily in glacial till landforms 

with loam or silt soils. 


Average potential recharge varied substantially across the area. For 1971-1990, it generally ranged from 0 

to 15 inches per year and for 2046-2065 it tended to range from 0-20 inches per year. The average 

differences (future minus historic) for each pixel were mostly in the range of -1 to +2 inches (Figure 12). 

The average potential recharge increased 0.54 inches from 7.81 to 8.35 inches for the entire area covered 

by the Park Falls HUC-6s (Table 3). This represents a 6.9 percent increase in potential groundwater 

recharge. While not large, this could have a significant effect over time on some groundwater dependent 

resources. 

Figure 10. Percentage of total and acid wetlands for portions of HUC-6 watersheds within the 

Chequamegon-Nicolet National Forest derived from ecological land type inventory mapping 



Wetland Vulnerability Rating 

All Acid Combined 

low (0-10%) low (0-10-30%) low (0-10-30%) moderate (5-10%) moderate 

moderate (>10-30%) high (>10%) high 

high (>30%) low (0-30%) moderate (5-10%) high 

high (>30%) high (>10%) very high 

Table 2. Wetland vulnerability ranking criteria for HUC-6 watersheds 

on the Chequamegon-Nicolet National Forest 

Figure 11. Relative vulnerability of wetlands to climate change for HUC-6 watersheds on the Chequamegon-Nicolet 

National Forest 

The small increase in potential groundwater recharge can be explained by the timing of groundwater 

recharge and projected changes in the climate of northern Wisconsin. In northern Wisconsin and 

throughout much of the Lake States, most groundwater recharge occurs in spring when there is excess soil 

moisture at the end of the snowmelt season and prior to the onset of summer (Boelter and Verry 1977). 

While the GCM projections for precipitation are generally less consistent than for temperature, they tend 



to show a small increase in precipitation during fall, winter, and spring for northern Wisconsin. This 

additional water, available at the time of year when evapotranspiration is low, will most likely go to 

satisfying soil moisture deficits and recharging groundwater. 

Both the absolute potential groundwater recharge and the difference for the two time periods varied by 

soil type. Highly permeable soils have greater potential recharge and showed a greater positive difference 

than heavy or peatland soils. Average potential recharge ranged from 13.5 inches for HSG A to 3.5 inches 

for HSG D (Table 3). HSGs A, B, C, and D had average increases of 1.3, 0.8, 0.7, and 0.0 inches, 

respectively (Table 3, Figure 13). HSGs are based on runoff potential when soils are thoroughly wet, 

considering texture, presence of impermeable layers, and depth to water table. HSG A soils have low 

runoff potential and consist primarily of sand and gravel. HSG B soils have moderately low runoff 

potential, consisting of mostly loamy sand and sandy loam textures. HSG C soils have moderately high 

runoff potential and finer textures such as loam, silt loam, sandy clay loam, clay loam and silty clay loam. 

Hydrologic 

Soil Group 

Area 

(acres) 

Avg. Annual Potential Recharge 

(inches) 

2046- 

2065 

1971- 

1990 


Mean 

Difference 

A 62,351 14.88 13.54 1.34 

B 96,384 11.51 10.75 0.76 

C 37,134 7.16 6.51 0.65 

D 116,218 3.47 3.51 -0.04 

Water 14,144 1.19 1.17 0.02 

Total 326,231 8.35 7.81 0.54 

Table 3. Summary of average annual potential groundwater recharge (inches) by hydrologic soil group for HUC-6 

watersheds on the Park Falls Unit of the Chequamegon-Nicolet NF 

HSG D soils have high runoff potential because of clayey textures, an impermeable layer within 20 

inches, or water table within 24 inches. Based on the results of the groundwater recharge modeling, HSG 

As were considered least vulnerable or most resilient to climate change impacts while HSG Ds were 

considered most vulnerable or least resilient. 

Because the response of potential groundwater recharge to the projected climate change varied by HSG, 

this information was used to estimate potential groundwater recharge and vulnerability to climate change 

for each HUC-6 on the Forest. An HSG index was developed for each HUC-6, based on the areaweighted 

proportion in each HSG, with A=1, B=2, C=3, and D=4. This index was used, along with the 

presence of surface water features, to classify the watersheds into four classes: groundwater recharge 

(HSG index2.837). Regression analysis was used to relate this index to the future and historic potential 

groundwater recharge for the HUC-6s on the Park Falls unit. These regression equations were then used 

to estimate and summarize potential historic and future recharge for all HUC-6s across the Forest.


Figure 12. Difference in potential groundwater recharge average 2046-2065 minus 1971-1990, Park Falls Unit, 

Chequamegon-Nicolet NF 



GW Recharge Difference (inches/yr) 

1.5 

1.25 

1 

0.75 

0.5 

0.25 

0 

-‐0.25 

A B C D Water 

Hydrologic Soil Group 

Figure 13. Average annual difference in potential groundwater recharge by hydrologic soil group 

These estimates need to be viewed cautiously; but in spite of these shortcomings, the regressions are 

strong, they are consistent with the modeled results by HSG, and the climatic differences across the Forest 

are not large. Therefore, they should provide reasonable estimates for the entire Forest until more 

comprehensive modeling can be conducted. 

The number, location, and relative vulnerability of the four HUC-6 classes are presented in Figure 14. 

The groundwater recharge watersheds have few or no streams, very high permeability, and are entirely 

groundwater recharge zones that were considered to be resilient or to have very low vulnerability to 

impacts from the projected climate change. There were 12 watersheds in this class; nine were split HUC- 

6s and three were complete HUC-6s. All were located on the Bayfield Peninsula. The estimated average 

annual future and historic potential groundwater recharge for these watersheds were 13.4 and 12.4 inches, 

respectively, resulting in an average increase of 1.0 inch (Table 4). These watersheds may provide the 

best opportunities on the Forest to implement adaptive management practices to respond to climate 

change for resources other than water. 

HUC-6 # of Est. Avg. Annual Potential 

Watershed HUC-6 Groundwater Recharge (inches) 

Class Watersheds 2046-2065 1971-1990 Difference 

Groundwater Recharge 12 13.4 12.4 1.0 

Groundwater Runoff 50 10.1 9.4 0.7 

Mixed Runoff 59 8.6 8.1 0.6 

Surface Runoff 37 7.0 6.6 0.4 

Table 4. Estimated Average Potential Groundwater Recharge for the historic (1971-1990) 

and future (2046-2065) time periods for HUC-6 watersheds on the Chequamegon-Nicolet NF 



Figure 14. Relative vulnerability of groundwater recharge to climate change for HUC-6 watersheds on 

Chequamegon-Nicolet NF 

Runoff from the groundwater watersheds is dominated by groundwater discharge and they were 

considered to have low vulnerability. There were 50 HUC-6s classified as groundwater runoff. They were 

located predominantly in outwash sands on Lakewood/Laona RD, northern Eagle River/Florence RD, and 

western Great Divide RD. The estimated average annual future and historic potential groundwater 

recharge for these watersheds was 10.1 and 9.4 inches, resulting in an average increase of 0.7 inches 

(Table 4). These watersheds are most likely to provide refugia for groundwater-dependent resources such 

as brook trout and other cold water stream fish. They may be an area to focus adaptive management for 

these resources. 

Runoff from mixed watersheds includes a combination of groundwater and surface water and these 

watersheds were considered to have moderate vulnerability. There were 59 mixed HUC-6s located on 

Park Falls units, eastern Great Divide RD, and Eagle River/Florence RD. The estimated average annual 

future and historic potential groundwater recharge for these watersheds was 8.6 and 8.1 inches, resulting 

in an average increase of 0.6 inches (Table 4). Some of these watersheds may have a few cold water 

streams in local areas where soil and topography provide adequate groundwater recharge and discharge 

and these streams may be vulnerable yet important potential refugia. 

Runoff from surface watersheds is dominated by surface runoff processes and these watersheds were 

considered to be most vulnerable. There were 37 watersheds classified as surface runoff; 30 were 



complete HUC-6s and seven were split watersheds. These were located in the moraines on the Medford 

unit, the clay plain along Lake Superior, the southern half of Park Falls unit, southwest portion of Eagle 

River/Florence RD, and central portion of Great Divide RD. The estimated average annual future and 

historic potential groundwater recharge for these watersheds was 7.0 and 6.6 inches, resulting in an 

average increase of 0.4 inches (Table 4). With a few exceptions, these watersheds will contain very few 

surface waters that are substantially fed by groundwater and these will be the most susceptible to climate 

change impacts. The exceptions are the split watersheds on the Bayfield Peninsula, which have low 

groundwater recharge themselves but many of whose main streams are heavily fed by groundwater from 

upslope groundwater recharge watersheds and an occasional isolated coldwater stream. 


While it is not possible at this time to predict changes in flood frequency and magnitude due to climate 

change, the WICCI downscaled projections provide sufficient evidence that the frequency and intensity of 

large precipitation events will increase and are likely to increase floods. The WICCI Stormwater Working 

Group reported that more frequent and severe flooding in rural areas are likely from the projected 

increases in rainfall and shifting precipitation patterns that favor more rain during periods of low 

evapotranspiration and high soil moisture which result in lower infiltration rates (Potter et al. 2010). 

Maintaining the current infrastructure, minimizing natural resource impacts, and reducing life cycle 

maintenance costs will logically require road crossing designs that will last at least 75 and preferably 100 

years. Structures installed in the near future must last until the late 21 st century and survive future climate 

changes. 

The CNNF 2004 Forest Plan revision included a guideline that all road and trail stream crossings be 

designed to pass the 100-yr flood (USDA Forest Service 2004). Since 2004, the CNNF has attempted to 

design all crossings to pass the 100-year flood with the headwater-to-depth (HW/D) ratio of less than 1 

(i.e., water level below the top of the culvert) to prevent pressurized flow or surcharging in the structure 

and to provide freeboard. In 2008, the US Forest Service published a guide for simulating stream 

channels at road and trail stream crossings to maintain or restore ecological connectivity (USDA Forest 

Service 2008). This design procedure also maximizes structure life and minimizes maintenance 

requirements. Using this guide, a structure width is selected that will allow the construction of a channel 

with bankfull width and stable banks, and a structure height is selected that will prevent pressurized flow 

and maintain sediment transport. 

In recent years, the CNNF has used two procedures to design road and trail stream crossings: no-slope 

with tailwater control, and stream simulation. Both procedures consider bankfull width and pass the 100yr 

flood with HW/D


Figure 15. Modeled 100 and 500 year flood water surface elevations for an 87”x63” pipe-‐arch culvert at Riley 

Creek and Forest Road 2161 with a minimum bankfull width of 7.0 feet 

Road and trail stream crossings inventoried on the CNNF were used to estimate crossing density for each 

HUC-6. Densities for each HUC-6 ranged from 0.0 to 1.83 crossings per square mile. Watersheds were 

rated for their vulnerability to infrastructure impacts based on the following crossing densities (mi/sq mi): 

very low, 0.0-0.15; low, 0.16-0.39; moderate, 0.40-0.83; and high, 0.84-1.83. Watersheds were rated for 

their vulnerability to increased floods based on the following HSG indices: very low, 1.049-1.816; low, 

1.862-2.422; moderate, 2.446-2.837; and high, 2.838-5.894. 

Combining the HSG and crossing density indices while giving the HSG index double weight resulted in 

26 HUC-6s rated very low, 50 low, 46 moderate, and 37 high (Figure 16). The most vulnerable HUC-6s 

have high runoff potential and high crossing density while the least vulnerable have the opposite 

characteristics. However, it is possible to adapt to potential increases in flood flows in all watersheds by 

sizing stream crossing structures to bankfull width or greater and conducting hydrologic and hydraulic 

analyses to ensure the 100-yr flood elevation is below the top of the culvert to provide freeboard for 

future increases in flood flows. Such sizing will also help to restore or maintain aquatic organism passage 

and channel morphology, reduce maintenance, and extend structure life. 



Figure 16. HUC-6 watershed vulnerability to infrastructure (stream crossing) impacts from climate change on the 


Stream Fishes 

Both cold water species, brook trout and mottled sculpin, are very vulnerable to all levels of warming but 

especially to moderate and major warming. The projected existing and future brook trout distributions are 

provided in Figure 17. Brook trout and mottled sculpin were predicted to decline by 81 and 76 percent, 

respectively, under moderate warming, and 100 and 90 percent under moderate warming (Table 5). These 

two species are fairly common in small- to medium-sized streams across the CNNF and brook trout are a 

popular sport fish. Such declines could have a dramatic effect on recreational fishing opportunities and 

cold water stream ecology. 

As a group, cool water species appear to be very vulnerable to moderate and major warming. They were 

predicted to decline by 15 to 98 percent under moderate warming and only two of these species, brook 

stickleback and northern hogsucker, were predicted to decline by less than 47 percent (Table 5). These 

eight species are very common and occur in a wide range of stream habitats across the Forest. Such 

declines could have a dramatic effect on the abundance and distribution of stream fishes and on stream 

ecology. 



Figure 17. Predicted distribution of brook trout (Salvelinus fontinalis), a cold water species, for current climate and 

moderate warming (increase air 3 deg C, water 2.4 deg C), Chequamegon-Nicolet NF 

With the exception of hornyhead chub, warm water species considered in this analysis were predicted to 

remain the same or expand habitat. Black crappie and stonecat were predicted to expand substantially on 

a percentage basis but because their existing habitat is very limited (4 and 1%, respectively, of total 

stream length), the absolute increase in habitat would be less dramatic (10 and 21%, respectively, of total 

stream length) (Table 5). 

Since all fish habitat used in this analysis was predicted from modeling, including habitat for the present 

climate, this data is most useful when viewed as an index of the relative magnitude and general pattern of 

species distribution changes in response to future warming scenarios. This modeled habitat has been used 

here to classify the vulnerability of individual HUC-6s but the results for any individual HUC-6 should be 

viewed carefully and the use of more detailed and site specific data should be considered. 

For cold water fish, there were 35 HUC-6s (22%) classified as having high vulnerability, 35 (22%) as 

moderately vulnerable, 37 (24%) as low vulnerability, and 51 (32%) as having very low vulnerability 

(Figure 18). For cool water fish, there were 40 HUC-6s (25%) classified as having high vulnerability, 40 

(25%) classified as moderately vulnerable, 39 (25%) classified as low vulnerability, and 39 (25%) 

classified as having very low vulnerability (Figure 19). 



Fish Species 

brook trout (Salvelinus 

fontinalis) 

brown trout (Salmo trutta) 

mottled sculpin (Cottus 

bairdii) 

blacknose dace (Rhinichthys 

obtusus) 

brook stickleback (Culaea 

inconstans) 

creek chub (Semotilus 

atromaculatus) 

longnose dace (Rhinichthys 

cataractae) 

northern hogsucker 

(Hypentelium nigricans) 

northern redbelly dace 

(Phoxinus eos) 

Thermal 

Class 

Sensitivit 

y Class 

Size 

Class 


Climate Warming Scenarios 

Current Climate Limited Warming Moderate Warming Major Warming 

Length 

(km) 

% of 

Total 

Length 

Length 

(km) 

% 

Change 

% of 

Total 

Length 

Length 

(km) 

% 

Change 

% of 

Total 

Change 

Length 

(km) 

% 

Change 

cold S H 3122 50 2743 -12 44 603 -81 10 0 -100 0 

cold S H 634 10 633 0 10 582 -8 9 289 -54 5 

cold S H 4700 76 2983 -37 48 1137 -76 18 448 -90 7 

cool T H 4927 79 4836 -2 78 1049 -79 17 613 -88 10 

cool T H 2913 47 2906 0 47 2467 -15 40 1200 -59 19 

cool T H 5244 85 4501 -14 73 1878 -64 30 1003 -81 16 

cool S M 2051 33 2045 0 33 728 -65 12 126 -94 2 

cool S R 1180 19 1143 -3 18 874 -26 14 183 -84 3 

cool S H 4877 79 4594 -6 74 82 -98 1 0 -100 0 

walleye (Sander vitreus) cool S R 289 5 283 -2 5 152 -47 2 0 -100 0 

white sucker (Catostomus 

commersonii) 

cool T U 3164 51 2836 -10 46 711 -78 11 158 -95 3 

black crappie (Pomoxis 

nigromaculatus) 

warm M R 222 4 534 141 9 1261 468 20 1261 468 20 

hornyhead chub (Nocomis 

biguttatus) 

warm S M 3211 52 3192 -1 51 679 -79 11 760 -76 12 

logperch (Percina caprodes) warm S R 1307 21 1159 -11 19 1086 -17 18 1407 8 23 

smallmouth bass 

(Micropterus dolomieu) 

warm S R 613 10 613 0 10 613 0 10 613 0 10 

stonecat (Noturus flavus) warm S M 55 1 334 507 5 590 973 10 633 1051 10 

Table 5. Summary of predicted fish habitat under three warming scenarios for HUC6 watersheds encompassing the Chequamegon-Nicolet National Forest (for 

sensitivity: S=sensitive, M=moderate, T=tolerant; for size class: H=headwater, M=mainstem, R=riverine, U=ubiquitous; findings based on Lyons et al. 2010) 

% of 

Total 

Length


Figure 18. Predicted vulnerability of 2 species of coldwater fish by 6th level watershed for moderate warming 

(3 deg C increase), for Chequamegon-Nicolet NF 



Figure 19. Predicted vulnerability of 8 species of coolwater fish by 6th level watershed for moderate warming 

3 deg C increase), for Chequamegon-Nicolet NF 


Based on the composite watershed vulnerability ratings, 11 HUC-6s were rated very low, 59 low, 64 

moderate, and 24 high (Figure 20). The watersheds with very low vulnerability were exclusively or 

predominantly groundwater recharge zones. These were rated very low because they support low 

densities of the water resource values (wetlands, stream crossings, cold and cool water stream fisheries). 

They also contain highly permeable soils, in which adverse effects to groundwater recharge from climate 

changes are least likely. The vulnerability of other watersheds depended on the combined occurrence of 

wetlands, runoff potential, road-stream crossing density and the presence of cold and cool water fisheries. 

As the occurrence of these attributes increased, so did overall watershed vulnerability to climate changes. 



Figure 20. HUC-6 vulnerability to climate change based on 5 attributes (groundwater, wetlands, cold and cool water 

stream fish, and infrastructure-culverts) for Chequamegon-Nicolet NF 

CONCLUSIONS 

Wetlands 

Hydrologic modeling of an upland-bog complex with PHIM for the Park Falls unit of the Chequamegon- 

Nicolet NF, using WICCI downscaled data for one location, just one GCM (GFDL-CM2.0) and one 

climate change scenario (A1B), indicates that bogs may be susceptible to climate change impacts. 

Average annual evapotranspiration would increase about 3.2 inches or 15 percent, runoff could decrease 

about 1.3 inches or roughly 25 percent with increases in spring and decreases in summer and fall, water 

levels in the bog would be 2-4.5 inches lower in summer and fall, and no-flow days would increase from 

about 4 to 23 percent of time. 

The PHIM modeling may have underestimated runoff, especially in spring, but the overall results seem to 

provide reasonable estimates of the potential impacts of climate change on bog hydrology. Based on the 

modeling, it was concluded that climate change poses some risk to Chequamegon-Nicolet National Forest 

wetlands, especially bogs. These risks include loss of wetland area, changes in wetland plant 

communities, and alteration of wetland processes such as water chemistry, peat accumulation, and 



geochemical cycling. These results were extrapolated to all HUC-6s on the Forest based on their 

percentage of total and acid wetland, and each watershed was placed into one of four classes representing 

its vulnerability to climate change impacts on wetlands. 


Results from soil water balance modeling for the Park Falls unit of the Chequamegon-Nicolet NF, using 

WICCI downscaled data for one location, just one GCM (GFDL-CM2.0), and one scenario (A1B), 

indicates potential groundwater recharge may increase about 7 percent in the future. While these are 

preliminary results, they indicate that groundwater recharge might be somewhat resilient to climate 

change impacts. 

Potential groundwater recharge and increases in recharge were related to hydrologic soil group with 

coarse textured soils having the highest potential average recharge (13.5 in/yr) and increase in recharge 

(1.4 in) and fine textured or peat soils having the least potential average recharge (3.5 in/yr) and increase 

in recharge (0.0 in). 

These results were extrapolated to all HUC-6s on the Forest and each watershed was placed into one of 

four classes representing its vulnerability or resilience to climate change impacts on potential 

groundwater recharge. 


The WICCI downscaled climate projections provide sufficient scientific evidence that the frequency and 

intensity of large precipitation events will increase and will likely increase floods. Indices of road-stream 

crossing density and runoff potential based on HSG were developed and used to classify the vulnerability 

of HUC-6s to impacts on infrastructure. The most vulnerable watersheds have high runoff potential and 

high stream crossing densities. For watersheds with low to moderate runoff potential, sizing stream 

crossing structures to channel bankfull width is an adaptive strategy that will most likely accommodate 

future increases in flood flows. And while hydrologic and hydraulic modeling should be conducted for all 

stream crossing designs, it is especially important for watersheds with very high runoff potential. In those 

cases, hydraulic modeling should be conducted to ensure structures pass the 100-yr flood, and preferably 

the 500-yr flood, with the HW/D


Changes to predicted available habitat for the five warm water fishes under a moderate warming scenario 

vary significantly. Two were predicted to increase by 468 to 973 percent two were predicted to remain 

about the same, and one was predicted to decline by 79 percent. 

These results indicate that cold and cool water fish on the CNNF are very vulnerable to moderate and 

major warming. Such warming could cause large declines in these fish, which could substantially impact 

stream ecology throughout the CNNF. 

The predicted fish distributions for the current climate and moderate warming were analyzed to determine 

the percent change in cold and cool water fish habitat in each HUC-6 on the CNNF. These results were 

used to place each watershed into one of four vulnerability classes. The most vulnerable HUC-6s are 

those predicted to contain a substantial amount of habitat under the current climate but which also had 

substantial declines in predicted habitat with moderate warming. The least vulnerable HUC-6s are 

primarily those with little or no predicted habitat given the existing climate. 

The increase of 0.8 o C for each 1.0 o C increase in air temperature used by Lyons et al. (2010) in their 

study was an oversimplification necessitated by the statewide study that did not take into account how 

groundwater input, land uses, or changes in flow might alter the response of streams to air temperature 

increases. 


Watersheds with very low composite vulnerability were exclusively or predominantly groundwater 

recharge zones. These were rated very low because they support low densities of the water resource 

values (wetlands, stream crossings, cold and cool water stream fisheries). They also contain highly 

permeable soils, in which adverse effects to groundwater recharge from climate changes are least likely. 

The vulnerability of other watersheds depended on the combined occurrence of wetlands, runoff potential, 

road-stream crossing density, and the presence of cold and cool water fisheries. As the occurrence of 

these attributes increased, so did overall watershed vulnerability to climate changes. 

RECOMMENDATIONS 

Wetlands 

There is a need to conduct much more comprehensive wetland modeling with downscaled data from 

additional GCMs, scenarios, and locations to verify and refine the preliminary results described above. 

Modeling should also be conducted for a variety of bogs with different wetland and contributing 

watershed areas. 

Other wetland types, including vernal ponds, fens, and weak fens, should be modeled and evaluated for 

their vulnerability to climate change. 

Existing mapping that includes wetland units, such as Wisconsin Wetland Inventory, WISCLAND and 

Forest Service stand inventory, is inadequate to fully evaluate the potential impacts of climate change on 

wetlands because it does not adequately characterize water source and flow regimes. In addition, this 

mapping frequently does not include vernal ponds, does not incorporate watershed divides through 

wetlands, and may have inaccuracies due to limited field verification. National Forest ecological land type 

inventory mapping provides the most accurate information, but is limited to areas within the National 

Forest boundary. Wetland inventories and mapping should be upgraded as soon as to solve these 

shortcomings and to allow more accurate determination of wetland vulnerability to climate change. 



If the above recommendations are completed, the results should be used to identify and develop more 

specific adaptations to minimize the impact of climate change on wetlands. 


More comprehensive soil water balance modeling should be conducted with downscaled data from 

additional GCMs, scenarios, and locations to verify and refine the preliminary results described above. 

These results could then be incorporated into groundwater flow modeling to predict effects on aquifers, 

groundwater flow paths, and surface waters dependent on groundwater flow. 

Once such groundwater modeling is completed, the results need to be evaluated with regard to potential 

effects on important groundwater-dependent resources such as cold water streams, wetland fens, 

groundwater-fed lakes, and water supply wells. 

Use the results from the above activities to identify and develop more specific actions to adapt to the 

impacts of climate change on watersheds and water resources. 


There is a need to conduct hydrologic modeling using the WICCI downscaled daily precipitation data and 

a variety of watershed conditions to more accurately determine potential increases in flood flows 

associated with the projected changes in future climate. The CNNF should support such work to the 

extent practicable. 

The CNNF should conduct additional analyses of culverts. The evaluation should determine where sizing 

to bankfull channel width will adequately adapt to climate change and also assess aquatic organism 

passage and channel morphology. 

The CNNF should continue to size stream crossing structures using stream simulation guidelines. 

Structures should be sized to at least match minimum bankfull width and pass the 100-year and preferably 

500-yr flood with the HW/D


Additional research should be conducted regarding the thermal requirements and tolerance of cool water 

fish, to better clarify their vulnerability to warming and potential management options. This work should 

be supported by the CNNF and US Forest Service. The CNNF stream segment classification system 

should be used to better identify existing cool water stream habitat. 

The CNNF should continue to (1) implement best management practices for water quality, (2) practice 

sound watershed management, and (3) restore streams (e.g. properly replace stream crossings that 

impound water or prevent fish passage, restore streams impacted by log drives, manage beaver in critical 

habitat) to improve their resilience to climate change impacts. 

The CNNF should also continue to monitor stream temperatures across a variety of stream types to (1), 

gather year round temperature data, (2) provide up-to-date data on current stream temperatures, (3) more 

accurately identify vulnerable streams, (4) establish trends in stream temperature, and (5) facilitate more 

accurate modeling of response to climate change. 

REFERENCES 

Boelter, D.H. and E.S. Verry. 1977. Peatland and Water in the northern Lake States. USDA Forest 

Service General Technical Report NC-31, North Central Forest Experiment Station, St. Paul, MN, 22 p. 

Brooks, K.N., S.Y. Lu and T.V.W. McAdams. 1995. User Manual for Peatland Hydrologic Impact 

Model (PHIM), Version 4. College of Natural Resources, University of Minnesota, St. Paul, MN, 55108, 

150 p. 

Dripps, W.R. and K.R. Bradbury, K.R.. 2007. A simple daily soil-water balance model for estimating 

the spatial and temporal distribution of groundwater recharge in temperate humid areas. J. of 

Hydrogeology 15: 433-444. 

Eggers, S.D. and D.M. Reed. 1987. Wetland Plants and Plant Communities of Minnesota and 

Wisconsin. US Army Corps of Engineers, St. Paul District, St. Paul, MN, 201 p. 

Guertin, D.P., P.K. Barten and K.N. Brooks. 1987. The Peatland Hydrologic Impact Model: 

Development and Testing. Nordic Hydrology 18, p. 79-100. 

Hawkinson, C.F. and E.S. Verry. 1975. Specific Conductance Identifies Perched and Ground Water 

Lakes. USDA Forest Service Research Paper NC-120, North Central Forest Experiment Station, St. Paul, 

MN, 5 p. 

Lyons, J., J. S. Stewart and M. Mitro. 2010. Predicted effects of climate warming on the distribution of 

50 stream fishes in Wisconsin, U.S.A. J. of Fish Biology 77:1867-1898. 

McAdams, T.V., K.N. Brooks and E.S. Verry. 1993. Modeling Water Table Response to Climate 

Change in a Northern Minnesota Peatland. In: Management of Irrigation and Drainage Systems 

Symposium, July 21-23, 1993, ASCE, Park City, Utah, p. 358-365. 

Urie, Dean. 1977. Groundwater differences on pine and hardwoods forests of the Udell Experimental 

Forest in Michigan. USDA Forest Service Research Paper NC-145, North Central Forest Experiment 

Station, St. Paul, MN, 12 p. 



USDA Forest Service. 2008. Stream Simulation: An Ecological Approach to Providing Passage for 

Aquatic Organisms at Road-Stream Crossings. USDA Forest Service, National Technology and 

Development Program, 7700 Transportation Management, 0877 1801-SDTDC, 646 p. 

USDA Forest Service. 2004. Chequamegon-Nicolet National Forests, 2004 Land and Resource 

Management Plan. Chequamegon-Nicolet National Forest, R9-CN-FP, 500 Hanson Lake Road, 

Rhinelander, WI 54501. 

US Geological Survey and US Department of Agriculture, Natural Resources Conservation Service. 

2009. Federal guidelines, requirements, and procedures for the national Watershed Boundary Dataset: 

U.S. Geological Survey Techniques and Methods 11–A3, 55 p. 

Walker, J.F. and W.R Krug. 2003. Flood-Frequency Characteristics of Wisconsin Streams. US 

Geological Survey, Water Resources Investigations Report 03-4250, 37 p. 

Westenbroek, Kelson, Dripps, Hunt and Bradbury. 2010. SWB-A modified Thornthwaite-Mather Soil 

Water Balance code for estimating groundwater recharge. US Geological Survey Techniques and 

Methods 6-A31, 60 p. 




Chugach National Forest 

March, 2012 

Prepared By: 

Ken Hodges 

Fisheries Biologist 

Cordova Ranger District, Chugach National Forest 

Cordova, Alaska 


SUMMARY 

This study was conducted as part of the USDA Forest Service Watershed Vulnerability Assessment Pilot 

Project. The goal of this study is to determine methods for assessing the vulnerability of aquatic resources 

to the predicted effects of climate change in the Chugach National Forest of southcentral Alaska. Many of 

the findings would also be applicable to coastal areas of southeast Alaska as well. 

The Chugach National Forest is somewhat exceptional in the National Forest system. Most of the Forest 

is undisturbed, with only 272 miles of road on 5.5 million acres, mainly state highways. There are no 

grazing allotments, no current commercial timber production to speak of, and limited active mineral 

extraction. Most of the Forest is managed for recreation and the conservation of fish and wildlife habitat. 

Climate change data were obtained from the University of Alaska, Fairbanks Scenarios Network for 

Alaska Planning program. Data are available online and some custom services were provided by the 

University. A review of the literature was conducted to determine how these changes are predicted to 

affect fish and wildlife, glaciers, and vegetation. 

Given that most of the watersheds in the Chugach are relatively pristine, ranking the vulnerability of all of 

the watersheds did not seem necessary. The large differences between ecosystem types were also not 

conducive to meaningful comparisons. Instead, two representative watersheds were selected for analysis: 

the Eyak Lake watershed near Cordova was chosen as representative of the coastal temperate rain forest 

ecosystem, and the Resurrection Creek watershed near Hope as more typical of the drier boreal forest of 

the Kenai Peninsula. Both watersheds are among the most developed on the Forest, although the overall 

disturbance may be considered low. 

Mean annual temperatures, precipitation, and days below freezing were developed for the watersheds by a 

Forest GIS specialist. Monthly data for Cordova and Hope, and other data are available online. Air 

temperature are predicted to increase in both areas, with summer temperatures increasing about 1.5 °C, 

but winter temperatures increasing about 4 °C. Precipitation is predicted to increase for all months in both 

watersheds, with a mean annual increase of 2 inches in the Resurrection Creek watershed and of 6 inches 

in the Eyak Lake watershed. All of these changes are well within the historic extremes. No predictions 

for extreme events in the future are available. 

Streamflow and water temperature data are limited in much of Alaska, and for the remote parts of the 

Chugach in particular. There are some stream gauge data for Resurrection Creek and Power Creek, which 

flows into Eyak Lake; however, the number of years of data are limited. I am unaware of VIC or other 

models that can be used to predict future flows with the available climate change data. Modeling flows is 

also complicated by conflicting factors. Snowpacks at lower elevations may be reduced by warmer 

temperatures in the fall and early spring, but this may be offset by higher precipitation and more snow at 

higher elevations. In addition, increased glacial melting may augment flows in late summer, which may 

compensate for an earlier melting of the snowpack – at least until the glaciers are gone. Given this 

complexity and limitations on the availability of modeling expertise, future conditions were judged 

qualitatively for each watershed. 

The assessment focused on the resource values in the watersheds, and particularly on the actions that 

could be taken to mitigate the predicted effects. Increased precipitation and the greater risk of rain-onsnow 

events make flooding and its effect on salmon habitat one of the greatest threats in both watersheds. 

Maintaining floodplain connectivity in the Eyak Lake watershed and restoring connectivity in the 

Resurrection Creek watershed are seen as the two most important mitigation measures to reduce the risks 

of salmon redd scour and other habitat damage. Increased erosion caused by higher precipitation, snow 


Chugach National Forest Watershed Vulnerability Assessment, Alaska Region (R10) 

avalanches, and exposure of glacial moraines could lead to higher bedload transport and channel shifts in 

depositional areas. This deposition, however, is seen as a natural response and does not pose risks to 

infrastructure or other values. Some forms of fish habitat enhancement that might be considered for 

mitigation, such as instream structures, may not be appropriate due to the potential channel instability. 

Managers also need to review existing restoration plans, road maintenance plans, and other work that 

already has been identified. Mitigation measures for the increased risk of fire in the Resurrection Creek 

watershed are already spelled out in the All Lands/All Hands program developed with other agencies and 

the Kenai Peninsula Borough. Fuel reduction goals, public education, and emergency preparedness 

measures are already lined out and are being implemented. Other entities, such as the Copper River 

Watershed Project in the Cordova area, have ongoing restoration programs, including the Million Dollar 

Eyak Lake project. Thus, Forest Service managers may have many opportunities for collaborative work. 

The greatest issue, however, may be the uncertainty as to how fish and wildlife species may respond to 

the effects of climate change. Salmon, in particular, are a key part of the ecosystem and the economy in 

Alaska. Unlike areas in the lower 48 states, coastal streams will have more, not less, water, and water 

temperatures will not rise enough for lethal effects to salmonids. Direct mortality is unlikely, but 

increased water temperatures could disrupt seasonal timing and life history cycles of both the fish and the 

food chains upon which they depend. If, for example, warmer water temperatures cause salmon eggs to 

mature more quickly, the fry could hatch too early in the season when no prey is available – unless the 

maturation of zooplankton and other organisms is temperature-dependent and increases as well. Without 

this basic knowledge, it is difficult to determine how the resources will be affected. 

There are a number of other biological questions, particularly whether species have the genetic/behavioral 

plasticity to adapt to changes. As an example, most salmon can have a wide range of spawning times, 

habitats, and life-history patterns. If eggs develop more quickly with warmer water, perhaps latespawning 

stocks will preserve the species. Perhaps the best mitigation is for land managers to maintain or 

restore diverse habitats and the genetic stocks that use them (something managers should be doing 

anyway). This is not to say populations will not be stressed, and population managers may well need to 

reduce harvests or take other actions as species adjust. 

To answer some of the biological questions, researchers from the Pacific Northwest Research Station and 

a number of universities are conducting studies in the Cordova area. Two current studies involve looking 

at differences in salmon and aquatic invertebrate life histories and timing, based on different temperature 

conditions across the Copper River Delta, including some sites in the Eyak Lake watershed. In these 

cases, physical locations are being used as a surrogate for the temperature changes that are predicted from 

climate change. Additional baseline data is also being collected on surface and groundwater temperatures, 

another major data gap. 

In summary, extensive climate data resources are available through the University of Alaska, Fairbanks, 

but limited historic data and models may hinder quantitative assessments. However, determining climate 

change trends, identifying resource values, and analyzing how those resources might be affected may be a 

sufficient start for determining future actions. In Alaska, where most areas are relatively pristine, it made 

more sense to focus on more developed watersheds to identify specific issues and actions. 

Much of the mitigation efforts that need to be done are actions that may already be planned or should be 

the normal plan of work. Stream projects that restore natural flows and functions may be the best way to 

protect fish habitat and reduce the risks of floods. Most Forests have conducted watershed assessments, 

road condition surveys, and fire management plans. The standards may need to be reviewed in light of 

predicted changes, such as increasing cross drainage or culvert sizes for roads, but most of the problems 

may already be identified. Lastly, a number of other government entities, agencies, community groups, 



and NGO’s may have existing programs or grants. This is the case even in the small fishing town of 

Cordova, Alaska, and the rural Kenai Peninsula. 

INTRODUCTION 

The Chugach National Forest is somewhat exceptional in the National Forest system. Most of the Forest 

is undisturbed, with only 272 miles of road on 5.5 million acres, 175 of which are state or Forest 

highways. No roads for timber harvest remain open. There are no grazing allotments, no current 

commercial timber production to speak of, and limited active mineral extraction. From 1985 to 1997, 

timber harvest averaged 2 million board ft/year, but this was due mostly to the salvage of beetle-killed 

spruce in the early 1990’s. By 1997, commercial harvest was no longer economically viable. 

The aquatic resource issues are limited as well. There are no threatened, endangered, or sensitive aquatic 

species unless one includes the Forest Service Alaska Region-designated sensitive dusky Canada goose 

(Branta canadensis occidentalis) that nests in the wetlands of the Copper River Delta. With small human 

population centers in the surrounding areas, limited industry, high precipitation, and no local agriculture, 

the demand for water is relatively low. There are, however, two diversions for hydroelectric power 

generation. The main aquatic resource issue is maintaining the high salmon productivity in the streams for 

the sport, commercial, and subsistence fisheries. Of particular importance are sockeye (Oncorhynchus 

nerka), coho (O. kisutch), chinook (O. tshawytscha), and pink (O. gorbuscha) salmon. 

The 2002 Forest Plan and its updates anticipate little development on Forest land except for tourismrelated 

projects such as the expansion of existing campgrounds, additional trails, and more recreation 

cabins. Water use and the amount of area affected by these activities would be relatively small. Adjacent 

landowners have not proposed major development projects. 

Most of the remaining FS management activities are related to fuel reduction or wildlife and fish habitat 

restoration and enhancement. Vegetation management for ungulate browse would affect the greatest 

amount of land with up to 10,000 acres treated with prescribed burns, cutting back mature shrubs, or other 

treatments. Fuel reduction would affect 4,000 acres over 10 years. Additional areas may be treated with 

prescribed fire for wildlife enhancement. 

Most of the necessary stream restoration work has been completed, with the exception of continued 

restoration of placer-mined areas along Resurrection Creek and Cooper Creek. The trend for fish habitat 

projects in the future will be elective enhancement projects on a small scale. 

A recent watershed condition classification study has been completed for the Chugach National Forest. Of 

the 275 sixth level watersheds, 268 were rated as Condition Class 1 (the best ranking), 7 as Class 2, and 

none as Class 3. Thus, most of the watersheds are intact and functioning properly. Large landscape 

disturbances from future development are not foreseen. With some exceptions, when managers examine 

the effects of climate change, they may find that there is little they can do to improve matters without 

altering natural conditions. 

ASSESSMENT OBJECTIVES 

This study was conducted as part of the USDA Forest Service Watershed Vulnerability Assessment Pilot 

Project. The purpose of this assessment is to provide land managers on the Chugach National Forest, and 

similar areas of Alaska, with a method of assessing the vulnerability of watersheds to the effects of 

predicted climate change. This entails the identification of the important aquatic resources or values, the 

type and degree of climate change, and the effects on the values. Most important, however, this 



assessment will stress the course of action that managers can take to mitigate the predicted negative 

effects. 

Realistically, there are a number of limitations on the analysis, particularly simple hydrologic data. Most 

of the Forest is accessible only by aircraft or boat, so data collection has generally been limited to projectspecific 

sites on a short-term basis. Since many of the watersheds have little historic or proposed human 

disturbance, data collection has not been a priority. 

I also assume that given the predicted climate changes for the area, undisturbed watersheds are best left 

alone. Predictions for coastal Alaska include increased precipitation, higher temperatures, and more 

intense storm events. While there may well be changes in stream flows, flow timing, or other effects, 

trying to “correct” those effects without altering other natural processes may be difficult. In addition, 

where there are no direct effects to infrastructure or threats to population centers, land managers may 

have higher priorities. 

Thus, instead of looking at all of the watersheds on the Forest and trying to rank their vulnerability, this 

assessment focuses on two of the more highly developed watersheds where more data are available, 

where a wider variety of restoration activity might occur, and that are representative of their ecological 

areas. These are the Eyak Lake watershed in a coastal rainforest ecosystem near Cordova, and the 

Resurrection Creek watershed in a relatively drier boreal forest setting on the Kenai Peninsula. 

Figure 1. The Chugach National Forest, its location in Alaska, and the two watersheds that were examined for this 

study 

Another limitation is that many of the biological effects are intuitively predictable – such as warmer water 

temperatures causing salmon eggs to develop and hatch sooner – but how these individual effects interact 

with other components of the ecosystem are unknown or cannot be quantified. Thus, there is a vast need 

for biological research that can help land managers reach decisions for on-the-ground mitigation 

activities. 



This assessment was made based on the conditions of the Chugach National Forest along the southcentral 

coast of Alaska, but it could be applicable to other areas in coastal Alaska, including southeast Alaska. 

The intent of focusing on just two watersheds is to have them serve as examples for land managers who 

may have watersheds with similar issues. 

METHODS 

The directions that participants in this pilot project were given included a number of practical steps. These 

included: 

• Describing the assessment areas, existing conditions, and the major water resources, or the waterrelated 

values or benefits in these areas. 

• Determining the anticipated climate change and its degree, using various predictive climate 

models. 

• Describing the predicted changes to hydrologic processes. 

• Determining the effects on water resources or values. 

• Describing the conditions that might amplify the changes and effects (stressors) or reduce them 

(buffers). 

• Determining the degree of watershed risk. 

• Describing how the findings might be applied to management activities at various geographic 

levels. 

The initial steps required consultations with area managers, literature searches (particularly of the gray 

literature), collecting historic temperature and precipitation data, and determining the availability of site 

specific data such as stream flows or water temperatures. 

The University of Alaska, Fairbanks (UAF), in collaboration with government agencies and nongovernmental 

organizations, conducts the Scenarios Network for Alaska Planning project (SNAP), which 

provides climate change data using a variety of Global Circulation Models (GCM) linked with historic 

Parameter-elevation Regressions on Independent Slope Models (PRISM) data. The resulting SNAP data 

can then make climate change predictions based on historic data that also take into account elevation, 

topographic facet, coastal proximity, slope, and distance from weather stations. This is particularly 

important in Alaska where there are large areas with few or no stations. 

There are ready-made maps with 2 km cells available online for temperature and precipitation, but the 

scale increments are somewhat coarse: 3 °C for temperatures close to freezing and 50 mm increments for 

precipitation. However, these maps are sufficient to determine overall trends and a rough estimate of the 

amount of change. Analysis requires downloading the data. 

For the initial efforts, UAF provided me with GIS layers of the Eyak Lake watershed where I could 

manipulate the scales to better detect freezing points and finer changes in precipitation. Since the 

elevations range from near sea level to 4,600 ft, the temperatures and precipitation vary significantly over 

short distances. The data were an average of the five GCM’s that best matched historical data. 

After the project was expanded to include Resurrection Creek, a GIS specialist for the Chugach National 

Forest downloaded and manipulated additional data available from SNAP for both watersheds. By using 

the raw data for each 2 km cell, the GIS specialist was able to average and obtain mean values for the 

watersheds as a whole. This was done for annual mean temperatures, annual mean precipitation, the 

freeze day, and the thaw day. The freeze and thaw days are extrapolated predictions of when the average 



daily temperatures are below or above freezing. Changes in the number of days between the freeze and 

thaw provide clues about changes in the annual hydrologic cycle, such as earlier snowmelt and runoff. 

Figure 2. An example of a GIS product developed from SNAP data. The 2 km cells were clipped to the 

watershed boundaries and the mean precipitation for the watershed was calculated. 

Monthly temperature and precipitation data were obtained from the SNAP community charts that provide 

predictions for selected towns. These data are an average of the five best-fitting GCM’s. As described on 

the website, “SNAP then scaled down outputs to the local level using data from Alaskan weather stations 

and PRISM, a model that accounts for land features such as slope, elevation, and proximity to coastlines.” 

(University of Alaska, Fairbanks 2011). The data are predictions for the 2 km grid square closest to the 

town. The data provided are derived from an average of five models (out of a total of 15) that best fit the 

historic data. Variability among the models is generally in the range of 0-4 °F and 0-0.7 inches for 

precipitation (ibid). 



Figure 3. An example of the community graphs provided by SNAP. The black bars show the amount of variation 

among the five models used for these projections. Graphs are also available for precipitation and with projections for 

low and high emissions scenarios as well. 

The predicted changes to hydrologic processes were only examined qualitatively. In part, this was due to 

the limited availability of personnel, hydrologic models for Alaska, and stream gauge data. Moreover, the 

available quantity of water does not appear to be a problem in Alaska as it is in other areas. The climate 

models call for increased precipitation in all months of the year and the relatively high average elevations 

of the watersheds would appear to buffer potential changes in snowmelt and runoff timing. 

To determine effects on water resources and values, I investigated the use of the NetWeaver 

knowledge-based decision support system. The appeal of this and similar systems is that they can 

incorporate empirical data as well as “expert opinion” in a logical transparent method. I thought this 

might be useful given the limited amount of available data for current and historic stream flows, water 

temperatures, and other parameters. It might also have been useful for comparing conditions across 

watersheds, since the system output is a numerical score measuring how “true” a certain proposition 

might be – for example, “Watershed X can sustain a viable coho salmon population.” 

The usefulness of this method, however, is limited by the complexity of the situation, how qualitative 

input is scaled (high, medium, low or numerically), and the confidence the experts have in making a 

rating or judgment. In short, this method did not prove to be practical and the analysis was not completed. 

My experience, however, provides a practical lesson for land managers that will be addressed in the 

assessment section. 

Further determination of the effects on aquatic resources and values, and the overall watershed risks, were 

made qualitatively, based on information in the literature, consideration of stressors and buffers, current 

investigations in the area, and personal communications. The issues are complex and there is a great deal 

of uncertainty, especially with the biological effects. These will be presented in the discussion. 



ASSESSMENT AREA DATA 

Assessment Areas 

• Eyak Lake watershed – coastal rainforest ecotype 

• Resurrection Creek watershed – drier boreal forest ecotype 

Of the 85 fifth level HUCs and 275 sixth level HUCs on the Forest, only a handful have significant 

development, infrastructure, or active land management. The two watersheds used for this assessment 

have the widest variety of development and aquatic resource values for their respective ecotypes and can 

serve as representative watersheds for many coastal Alaska areas. The Eyak Lake watershed contains part 

of Cordova, a small city of 2,440 people. The town of Hope (182 residents) is adjacent to the lower part 

of the Resurrection Creek watershed. 

Eyak Lake Resurrection Creek 

Area (acres) 27,748 103,215 

HUC Level Two 6 th levels 5 th level w/ three 6 th 

National Forest Land 25,554 (92.1%) 100,839 (97.3%) 

Mean Annual Air Temperature 

July 

January 

5.3 °C 

12.5 

- 4.1 

Water Temperature Summer surface Eyak 

Lake 5.5 - 14.5 °C 

Power Creek mean 6.1°C 

annual 3.0 - 8.4 °C 

Mean Discharge (cfs) 

387 (ungauged) 

10-yr Flood 

8,700 


2.6 °C 

14.1 

- 7.2 


Mean 8.4 °C 

Range 5.5 -12.0 °C 

275 

2,400 

Mean Precipitation (inches) 130.25 22.15 

Lake/Pond Area (acres) 2,400 (8.6%) 80.5 (0.07%) 

Road Density - mile/mile 2 (total) 0.36 (25.1) 0.14 (35.1) 

Residential/Commercial Area (acres) 205.3 (0.7%) 53.1 (0.05%) 

Area Unvegetated Rock (acres) (%) 5,405 (19.7%) 11,391 (11.0%) 

Area Icefields/Glacier (acres) (%) 3,256 (11.7%) 245 (0.2%) 

Area > 70% Slope (acres) (%) 6,332 (23.1%) 7,660 (7.4%) 

Avalanche Area (58-173% slope) (acres) (%) 12,001 (43.7%) 20,952 (20.3%) 

Area > 500 ft Elevation (acres) (%) 20.553 (74.0%) 100,324 (97.2%) 

Fire, Including Prescribed Burns (acres) (%) 0 9,400 (9.1%) 

Mining Disturbance (acres) (%) 0 2,560 (2.5%) 

Trails (miles) 13.1 16.8 

Recreation Sites (cabins, camps, day use) 5 14 

Table 1. Current conditions for the Eyak Lake and Resurrection Creek watersheds. 

Assessment Area Climate Change Predictions 

The general predictions for both the Eyak Lake watershed and the Resurrection Creek watershed call for 

increased temperatures. Annual mean temperatures are predicted to increase 1.7 to 1.9 °C for the two 

watersheds under both the A1B and B1 scenarios (Table 2). However, winter temperatures are predicted 

to increase much more than summer temperatures. Monthly data for the entire watersheds were not 

readily available, but the January temperatures for the towns of Cordova and Hope are predicted to rise 

3.4 to 3.7 °C, and the July temperatures 1.5 to 1.8 °C.



Air Temperature °C 

Annual Mean Annual Mean 

A1B Scenario B1 Scenario A1B Scenario B1 Scenario 

2000-2009 4.3 3.5 1.4 0.5 

2020-2029 5.1 4.4 2.1 2.1 

2050-2059 6.1 5.3 3.1 2.4 

Cordova January Hope January 

2001-2010 -1.4 -3.3 -6.1 -8.2 

2031-2040 -0.2 0.7 -5.2 -4.2 

2061-2070 2.3 0.2 -2.3 -4.7 

Cordova July Hope July 

2001-2010 14.1 13.1 15.1 14.1 

2031-2040 14.3 14.4 15.2 15.4 

2061-2070 15.6 14.9 16.7 15.9 

Table 2. Predicted changes for annual mean air temperatures for the Eyak Lake and Resurrection Creek watersheds 

as a whole, and selected monthly temperatures for the towns of Cordova and Hope. 

Annual mean precipitation is generally predicted to increase, although the total amounts are quite 

different for the two watersheds. As shown in Table 3, the increase in the Eyak Lake watershed as a 

whole is predicted to be as much as 6.7 inches under the A1B scenario, while the Resurrection Creek 

watershed may see an increase of 3.1 inches. The data for the driest and wettest months for Cordova and 

Hope were taken from the SNAP community charts (2011), and generally show small increases over time. 

Unlike the other trends, the prediction for June 2061-2070 shows a slight decrease, but given the 

variability among the models used for the prediction (University of Alaska, Fairbanks 2011), this is 

probably not significant. 

It should also be mentioned that the historic annual precipitation levels are highly variable for the 

Cordova area. The Cordova airport weather station, which is about 10 km from the Eyak Lake watershed, 

has an annual mean of 96.26 inches, but a historic range of 54.41 to 139.34 inches. Thus, while an 

average annual increase of six inches will lead to higher flows and presumably more extreme events, the 

watershed already experiences extreme changes. This makes it difficult to determine how, or how much, 

geophysical and biological conditions will be affected. 

From 1979 to 1995, a low-elevation station near Hope had a precipitation range of 15.19 to 31.30 inches, 

with a mean of 22.15 (Kalli and Blanchet 2001). The predicted amounts for the entire watershed are 

higher as shown below, but the predicted changes still appear well within the historic range. 


Precipitation Inches 

Annual Mean Annual Mean 


2000-2009 177.2 176.9 34.5 38.0 

2020-2029 179.6 178.3 35.8 38.1 

2050-2059 183.9 179.0 37.6 39.8 

Cordova June Hope May 

2001-2010 7.81 7.52 0.85 0.89 

2031-2040 7.86 7.50 1.05 0.96 

2061-2070 7.70 7.69 1.14 1.01 

Cordova October Hope September 

2001-2010 21.13 20.83 3.52 3.73 

2031-2040 21.92 21.19 3.68 3.50 

2061-2070 22.10 21.23 4.47 3.97 



Table 3. Predicted changes for annual mean precipitation in inches for the Eyak Lake and Resurrection Creek 

watersheds as a whole, and selected monthly totals for the towns of Cordova and Hope. 

Using data provided by SNAP, our GIS specialist determined the mean ordinal freeze and thaw dates, and 

from this we could derive the number of days where the average daily temperature was below freezing. 

The results do not appear to be consistent with other findings, since the B1 predictions suggest that 

conditions would be much warmer (later freeze and earlier thaw) than for the A1B scenario. 

Days Mean Temp < 0 °C Eyak Lake Resurrection Creek 


2000-2009 91 76 165 166 

2030-2039 62 27 152 99 

2060-2069 21 34 118 74 

Table 4. Predicted changes for the number of days below freezing for the Eyak Lake and Resurrection Creek 

watersheds as a whole 

General Area Climate Change Predictions 

Two other factors have the potential to exacerbate the effects of temperature and precipitation change: the 

predicted increase in extreme weather events and the accelerated melting of glaciers. Most sources agree 

about the trends, but it is difficult to predict the magnitude of these changes. It appears likely, though, that 

both will lead to increased stream flows, changes in sediment transport, and the potential for flooding. 

Extreme Weather Events 

Specific predictions for extreme weather events in the project area are not available. For high northern 

latitudes, however, Sillman and Roeckner (2008) state that there will be significant increases in the 

maximum and minimum temperatures and the amounts of precipitation for 5-day events and the 95 th 

percentile of wet days. They conclude that northern areas that have wet climates under the current 

conditions will become substantially wetter by the end of the 21 st century. 

Glacial Melting 

Site-specific conditions can greatly affect glacier formation or melting (Boggild et al. 1994, Dowdeswell 

et al. 1997, Arendt et al. 2010). Boggild et al. (1994) suggested that increased precipitation could add to 

glacial mass in Greenland, where there is an extensive higher-elevation land mass. Coastal Alaska has a 

number of high elevation glaciers as well. Topography, slope aspect, and local weather conditions, such 

as wind, can also affect accumulation of ice (Boggild et al. 1994). On the other hand, Crisitiello et al. 

(2010) found that the mass balance of two southeast Alaska glaciers has declined and has been correlated 

with temperature but not with precipitation. This suggests that increasing precipitation may not 

compensate for increased glacial melting. 

Closer to the project area, however, recent data indicate that most Alaskan glaciers are losing mass. 

Arendt et al. (2010) reported most of the 67 Alaskan glaciers surveyed during an early period from the 

1950’s to the mid-1990’s, and 28 glaciers resurveyed from the mid 1990’s to 2001, had thinned. Less than 

5% of the glaciers in the study had thickened, and most of these were tidewater glaciers where the melting 

of the toe of the glacier may have triggered other responses such as glacial surges (Arendt et al. 2010). 

They also found that during the latter period, when glaciers were resurveyed, the thinning rate was twice 

that of the earlier period. 



The glaciers that had thinned included the Scott Glacier that is adjacent to the Eyak Lake watershed and 

the Wolverine Glacier and Harding Icefield complex on the Kenai Peninsula, about 50 miles from the 

Resurrection Creek watershed. Thus, although we have no data for glacier or icefield melting within these 

watersheds, and cannot predict how increased precipitation might affect the mass balance, the recent 

trends suggest that there will be an eventual loss of glaciers. 

The predicted effects of glaciers melting are varied. Haufler et al. (2010) suggest that flows may initially 

increase with the additional meltwater, but that over time, the reduction in melting ice may cause streams 

to disappear. The higher meltwater flows may also erode unconsolidated glacial moraines, especially 

where glaciers have recently receded and the moraines are not vegetated. This could lead to increased 

sediment transport and eventual deposition in downstream areas. 

ASSESSMENT AREA RESULTS AND FINDINGS 

Eyak Lake Watershed 

Figure 4. Eyak Lake Watershed. The downstream delineation of the watershed is somewhat arbitrary as it is joined 

by the glacial Scott River to form an interwoven complex of channels before entering the Gulf of Alaska to the 

south. 

Area Description 

The primary reason for selecting the Eyak Lake watershed is that it is the most developed watershed of 

the eastern two-thirds of the Chugach National Forest. It also has the greatest range of aquatic resource 

values that might be affected by climate change. Predicted increases in air temperature, precipitation, and 

extreme weather events could result in damage to salmonid habitat, changes to salmonid life histories, 



damage to infrastructure, changes in hydroelectric production, and increased flooding of residential and 

commercial areas. Flooding is already a problem on a 5- to 10- year basis. 

The Eyak Lake watershed also has an active restoration program in place that can provide ideas and 

examples for land managers in other areas. The Copper River Watershed Project (CRWP), a local nonprofit 

group, has led a watershed restoration planning team with representatives from the Alaska 

Department of Fish and Game (ADFG), the Native Village of Eyak, the City of Cordova, the USDA 

Forest Service, the Prince William Sound Science Center, and other agencies, organizations, and 

individuals. Some of the completed activities and proposed projects will be discussed in the section on 

recommendations. 

Watershed Values 

• Large sockeye and coho salmon runs, average annual index counts 19,000 and 10,000, 

respectively (Botz et al. 2009). Extensive rearing areas in a shallow lake. Spawning habitat along 

the shore and in tributaries. No current population concerns exist. 

• Salmon populations support commercial, sport, and subsistence fisheries. Other salmonids 

provide sport fishing. 

• Residential and light industrial areas around lake and on floodplain downstream from the lake. 

This floodplain currently experiences flooding every 5 to 10 years. 

• Hydroelectric power generation on Power Creek. 

• Floatplanes use lake, small wheeled planes land on airstrip along lake. 

• Backup water supply for city of 2,000 people, three salmon processors/canneries. 

• Wildlife viewing – bears, waterfowl, and fish. 

Ecological Triggers and Thresholds for these Values 

• Water temperatures: 12-15 °C is optimal. 25 °C is lethal for salmonids. 

• A minimum of 5 cfs is needed in the Power Creek area bypassed by the hydroelectric diversion. 

The plant can utilize up to 320 cfs. (Mean creek flow 50-500 cfs.) 

• Flooding occurs when lake rises approximately 5-6 ft. 

• Floods at an unknown velocity may mobilize spawning gravels, destroy eggs. 

Data Available, Data Needs 

• Power Creek (main tributary) gauge data 1948-1995. Currently, the required 5 cfs flow is 

maintained mechanically and is monitored by the electric company. The total flow above that 

level is not monitored. 

• The Prince William Sound Science Center and CRWP have done some monitoring of water 

quality in Eyak Lake for the past few years. Eventually they will have more consistent data for 

temperature, dissolved oxygen, and other parameters. There is only limited water quality data for 

Power Creek. 

• Historic precipitation and air temperature data are available from gauges at the Cordova airport 

and a station in town. These are not in the Eyak watershed but are geographically close. 

• Need to correlate precipitation, cfs in Power Creek, with flood events in lake and Eyak River. No 

lake height data available, but a gauge has been installed on a downstream bridge this past year. 

• Preliminary groundwater temperatures for one winter taken by Gordon Reeves and Steven 

Wondzell, USDA PNW Research Station. 

• SNAP program conducted by the UAF has predictions for temperatures, precipitation, and 

freeze/thaw dates at a 2km scale. This was calculated with PRISM and five climate models. On- 



line maps and bar graphs are available for Alaska communities. Raw data are available for use 

with GIS. 

• Total salmon spawning area not known. 

Sensitivity 

The sensitivity of the Eyak Lake watershed is due mainly to natural conditions –steep slopes, shallow 

lake, and high precipitation. Human activities, such as road building and other development, have been 

relatively minimal but there are stressors that might affect watershed’s ability to respond to the predicted 

increases in temperature and precipitation. 

• The mean high elevation of the watershed makes the watershed less sensitive to the effects of 

higher temperatures on the glaciers and snowpack. However, the current storm patterns from the 

relatively warm ocean already cause frequent rain-on-snow events. These are likely to increase 

and occur at higher elevations. 

• The relatively high percentage of area covered by glaciers and icefields makes the watershed 

more sensitive to the effects of melting glaciers: increased flows and erosion of glacial moraine. 

• Stream temperatures could rise with predicted changes in air temperatures but should be well 

within the suitable range for salmonids. Power Creek temperatures were no more than 10 °C at a 

downstream location (Sea-Run Fisheries 2006) and should rise no more than the predicted 2 to 3 

°C air increase. All streams are relatively steep and short, so there is little opportunity for streams 

to warm. 

• Water temperatures for most of Eyak Lake are dominated by stream and groundwater flows. 

Summer surface temperatures are generally less than 13.5 °C and data at two sites suggest there is 

a thermocline at about 1 m (Crawford 2010). 

• Parts of Eyak Lake could be sensitive to higher water temperatures. The west arm of the lake has 

less circulation, is less than 3 m deep, and currently has recorded surface temperatures of 15 °C 

(the top end of the optimal range for salmonids). 

• Watershed is naturally flashy due to 19.7% being unvegetated, 23.1% having steep slopes 

(>70%), along with thin soils, high precipitation, and long duration of storm events. 

• Flooding already occurs in residential areas along Eyak River and Eyak Lake. Floods have 

occurred in 1983, 1985, 1986, 1995, 2004, and 2006. 

• Hydroelectric power generation is sensitive to flows in Power Creek, which are at a minimum in 

winter when precipitation is bound as ice and snow. Higher precipitation, warmer temperatures, 

and rising snowline could increase winter power generation. 

• Salmon spawning in the lake and smaller tributaries not sensitive to redd displacement by floods. 

Those fish spawning in the main channel of Power Creek may be susceptible to substrate 

mobilization. 

• The risk of avalanches could increase as warmer temperatures create more frequent wet, heavy 

snowpacks. There are a high percentage of avalanche-prone slopes of 58 -173%. There have been 

three fatal incidents in past 15 years. 

• Landslides. High precipitation, long steep slopes characteristic of glacial U-shaped valleys, thin 

soils, underlying bedrock, and glacial till increase propensity for landslides. 

Stressors 

Stressors from residential development include hydrocarbon input to the lake from the streets and snow 

dumping, nutrient input from fertilizer and leach fields, minor sediment input from unpaved roads, and 

runoff from two subdivisions. Low levels of hydrocarbons have been detected in water samples, but the 

overall effects of oil and the other stressors have not been quantified. 



Road density is low (0.12 km/km 2 ) with a total of 25.1 km of road in the watershed. One publication 

(NOAA 1996) rates this as well within the level of a properly functioning watershed (< 1.2 km/km 2 ). 

Residential development and roads along the lake have reduced lakeside vegetation. Invertebrates that 

fall from terrestrial vegetation make up a large part of the diet for juvenile coho salmon and this dietary 

input will be reduced. The effect on the water temperature of the lake as a whole is probably negligible, 

given the large areas far from shore and stream inputs. However, cooler, shallow shoreside areas, 

preferred by juvenile coho salmon for habit and rearing habitat, are reduced. 

Salmon spawning area in Eyak Lake has been reduced from 63,011 m 2 (Professional Fishery Consultants 

1985) to a currently unknown amount. This is a result of housing development, construction of a water 

treatment plant, and sedimentation from roads in one area. 

An unknown amount of salmon spawning area exists in the creeks. Culverts have reduced salmon 

spawning by several hundred meters, but the overall percentage of spawning area is minimal. There are 

perched culverts that do not prevent access to usable habitat, but do eliminate intergravel flows in alluvial 

fans in the lake that could be used for sockeye salmon spawning. 

Cutthroat trout spawning area has been reduced by 35% due to culverts, houses, and roads covering 

potential spawning areas (Hodges et al. 1995). 

Another possible stressor is the reduction in the number of returning salmon due to the commercial, sport, 

and subsistence harvests. This harvest not only reduces the number of spawning fish, but also the 

availability of salmon for predators and the amount of nutrients provided by the carcasses for organisms 

throughout the food chain. A greater abundance of nutrients might help populations stressed by climate 

change in the future. 

There are anecdotal reports that there used to be more sockeye salmon early in the season, with the first 

fish reaching the spawning areas in May. This could be an effect of the variability of run sizes. The 

ADFG generally has the first commercial fishing opener on May 15. There is a need to carefully manage 

the early part of run to maintain the full genetic diversity. 

Water use is not seen as a stressor. The only water diversions are for backup municipal water use and for 

hydroelectric power generation. However, the backup municipal water use is infrequent and the water 

used for power generation is returned to the channel upstream of fish habitat, so there is minimal effect. 

Trends 

The population of Cordova has declined from 2500 residents in 2000 to 2240 residents in 2009. The use 

of migrant non-resident labor at canneries, decreased government employment, and the lack of other 

resource jobs are likely to keep the population and development from growing. 

Almost all of the areas suitable for housing lots and roads in the watershed have already been utilized. 

Runoff from recently constructed roads and building lots should decrease as raw areas revegetate. The 

opening of 50 or more residential lots outside of the Eyak watershed will reduce development pressure. 

Overall, there have been no detectable trends for the salmon populations. The commercial salmon fishery 

is managed well, and minimum escapements in the watershed have been maintained. Population levels 

generally follow changes of weather patterns associated with the Pacific Decadal Oscillation and the El 

Nino and La Nina patterns (Chittenden et al. 2009). The sport fishery is not managed closely, but the 



harvest is still a small percentage of the commercial fishery (Lang 2010). Recreation and subsistence 

harvest are likely to grow, but data are lacking. 

Exposure/Risks 

Hydrologic/Geomorphic 

Assessing risk is difficult in the Eyak Lake watershed because the weather conditions are highly variable 

already. At the nearby Cordova airport, the mean annual precipitation from 1949 to 2004 is 96 inches, but 

the extremes have ranged from 54 to 139 inches (139 being 45% above normal). Thus, predictions that 

the mean precipitation in the rainier Eyak Lake watershed will increase 3% from 177 to 184 inches do not 

give a clear indication of how that will affect the hydrologic or geomorphic conditions. Such an increase 

is well within what might be considered normal. 

The significant changes are most likely to come from the extreme events, which are predicted to increase 

and intensify, but aren’t readily quantified. Mass wasting from snow avalanches is likely to increase but 

predicting such events is also not feasible. Thus, exposure and risk may be best discussed in general 

terms. 

The predicted increases in temperature and precipitation are likely to result in higher streamflows 

throughout the year, more frequent rain-on-snow events in the fall and spring, and changes in the timing 

of peak spring flows as the snowpack melts earlier. The predicted increase of extreme weather events, 

including increased storm duration and intensity, will also lead to greater streamflows. Glacial melting is 

expected to continue or accelerate, adding to flows in the summer, which could compensate for the 

reduction in flows from an earlier snowmelt. 

Geomorphically, these changes are likely to lead to increased snow avalanches, landslides, and other 

erosive processes. Temperatures changing between freezing and thawing at the lower elevations will be 

especially conducive to increasing snow avalanche danger. Many avalanche and landslide areas transport 

material directly to Power Creek or Eyak Lake itself, adding to the bedload. Exposed glacial moraines 

will be subject to erosion and transport by meltwaters. 

The increased bedload material will be deposited in the Power Creek delta at the head of Eyak Lake, and 

at Middle Arm and other smaller alluvial deposition areas around the lake. As with many deltas and 

glacial outwashes, stream channels will fill and shift. The Power Creek delta will most likely extend 

farther into the lake. 

The main consequence of the hydrologic and geomorphic changes will be the increased risk of flooding, 

especially in the subdivision just downstream from the outlet of Eyak Lake. Prolonged storm events in the 

fall have caused flooding in this area a few times every decade and this is only likely to increase with 

more precipitation and extreme events. Despite past flooding, development has continued on this 

floodplain due to the general scarcity of level land on which to build and its location beyond the city 

zoning areas. 

The other exacerbating factor is that flows from the glacial Scott River in the adjacent watershed can spill 

over into Eyak River, about 1/2 mile downstream from the development. As the Scott River deposits 

sediment into Eyak River, the Eyak channel’s ability to drain its watershed is reduced, resulting in 

increased flooding (Blanchet 1983, Hitch 1995). Similar increases in flows, bedload transport, and 

channel shifting in the Scott River are thus likely to affect Eyak River as well. 



Flooding will also affect Power Creek Road as it crosses the delta floodplain. Flooding already occurs 

every few years, but the flows are not sufficient to severely damage the dirt road, nor does the minimal 

amount of traffic seem to justify upgrading the road. Increased flows and a shift of the main channel, 

however, could cut off access to the hydroelectric plant until waters subside. 

One positive effect of the hydrologic changes may be the increased production of hydroelectric power at 

Power Creek. The plant is a run-of-the-river facility with no reservoir, so when winter precipitation falls 

as snow, and the river drops below 320 cfs, power generation is reduced. At the present time, maximum 

generation is reduced from late October to mid-May and severely limited from late November to April. 

The number of days with the mean temperature below freezing is predicted to decline dramatically, with 

precipitation falling as rain later into the fall and earlier in the spring. Thus, the period of higher power 

generation would be extended. Since the water use capacity of the turbines is well below the summer 

flows, and summer precipitation is predicted to increase, the smaller snowpack and summer runoff should 

still be sufficient to run the turbines at maximum capacity. 

Because there is no reservoir, additional bedload from increased erosion should not be a problem. There is 

a low dam with an inflatable bladder that can be deflated to allow accumulated sediment to be flushed 

from behind the wall and pass downstream. 

Biological Exposure/Risks 

In the western lower 48 states, the main concerns for aquatic organisms are high water temperatures and 

low flows that can have direct lethal effects. In coastal Alaska where precipitation will increase and water 

temperatures will be higher but still relatively low (Bryant 2009), the effects of climate change could be 

more subtle, but serious nonetheless. 

Water Temperature 

Water temperatures are expected to rise, but since existing stream temperatures in the Eyak watershed are 

cool, increases would not be lethal or even beyond the optimum levels for salmonids, the organisms of 

primary concern. Current lake temperatures are somewhat warmer, but Crawford (2010) shows that most 

of the lake temperatures are influenced by the streams, except for the shallow west end. Even there, 

surface temperatures in the summer have been moderate. If summer water temperatures increase about the 

same as the predicted air temperatures (1.5 to 1.8 °C), the temperatures would still be within or close to 

the optimal range. Thermal refugia would also be available near the mouths of some small creeks or in 

deeper waters. 

Increased water temperatures are more likely to have an effect on the egg and larval stages of fish and 

aquatic invertebrates. As is clear from fish hatchery experience (Piper et al. 1982), higher water 

temperatures accelerate the development of eggs and hatching. Based on a model by McCullough (1999), 

a 1 °C increase in water temperature could cause coho salmon fry to emerge about 10-20 days earlier in 

the Eyak area. If prey organisms do not follow the same pattern of earlier growth, the newly emerged fry 

may lack food resources (Bryant 2009). 

Such a scenario is described by Winder and Schindler (2004) where a species of zooplankton that 

emerged according to photoperiod length was at a disadvantage compared to a species that hatched by 

temperature. Unfortunately for sockeye salmon fry, their preferred prey species is the photoperiod 

dependent species, which may have significant effects in the future. Hypothetically, similar disruptions 

could occur with aquatic insect life cycles and the avian species dependent on them (McClure, et al, 

2011). It is not known whether similar scenarios may occur in the Eyak Lake area because the specific 



species in the food chain and their life histories have not been studied. This lack of information makes it 

difficult to assess the full effects of climate change. 

Another concern is that increased metabolic rates for juvenile fish in warmer water may result in reduced 

growth as a greater share of energy is expended for body processes when there is no increase in food 

availability (Bryant 2009). Smaller size is linked to higher predation rates. If fish have lower fat reserves 

going into the winter, winter survival rates will be a concern because food is less available then. Another 

research need is to determine whether food is a limiting factor or whether greater primary production 

from warmer temperatures and a longer growing season may lead to greater resources at higher trophic 

levels. 

Water Quantity 

Water quantity is generally not a concern, as increased precipitation throughout the year (in addition to 

the high current levels) should help to maintain flows in small streams. There is, however, some 

uncertainty about the degree to which warmer winter temperatures will affect the snowpack. Winter 

temperatures at sea level are expected to remain close to freezing until the middle of the century, but the 

number of days below freezing will decrease, and more precipitation is expected to fall as rain at the 

lower elevations. The question is whether the increased winter precipitation at the high elevations could 

offset this loss of snow and maintain the snowpack and, in turn, summer flows. 

The opposite concern is that flows may be too great. With increased precipitation, more frequent rain-onsnow 

events, and more extreme storm events, high streamflows in the fall could mobilize gravels in 

salmon spawning areas, displacing and killing the eggs in the redds. Material from landslides, triggered 

by extreme precipitation, could scour spawning beds or be carried by high flows and deposited on redds 

(Bryant 2009). Fine sediment deposition can not only smother salmon eggs; the deposition can cause 

greater and deeper scouring (Montgomery et al. 1996), dislodging eggs that might have been buried at a 

safe depth under other conditions. 

These risks might also be increased because warmer temperatures could extend the flood-prone season 

later into the year. Currently, by late October, most precipitation at higher elevations is falling as snow, 

and streamflows drop. The somewhat late spawning run of coho salmon in the main channel of Power 

Creek, which lasts into December, could be a local adaptation to avoid the risk of redd scour 

(Montgomery et al. 1999). However, the benefits of late spawning are negated if heavy rain or rain-onsnow 

events occur later in the year. 

Overall, however, the risks to spawning are buffered by the variety of spawning habitats used by 

salmonids. Sockeye salmon spawning in the lake is not subjected to scouring, although a large sediment 

flux or landslide could bury some areas. Much of the spawning of coho salmon and sockeye salmon 

occurs in the smaller, side channels of the Power Creek delta or in other tributaries that are not subject to 

high flows. Cutthroat trout spawning areas are almost all in small tributary streams (Hodges et al. 1995). 

Montgomery et al. (1999) and Tonina and McKean (2010) also stress that the channel type where 

spawning occurs influences the risk of redd scour. Steeper-gradient confined channels are naturally more 

prone to scouring, whereas less-confined channels allow flows and energy to be dispersed. In the case of 

the Eyak Lake watershed, most of the salmon stream spawning occurs on poorly controlled alluvial fans 

and in the Power Creek delta complex. As Tonina and McKean (2010) state: 

Our analyses showed that such unconfined low-gradient streams have not a great danger of extensive bed 

mobility, even at high flows. Consequently, in this landscape, alterations in flood timing due to climate 



change are unlikely to decrease the success rate of salmonid egg incubation by the mechanism of increased 

channel bed scour. 

Thus, salmon spawning in the watershed may be less sensitive to scour even with the predicted increases 

in flows, but this depends on maintaining floodplain connectivity. While it may seem appealing to elevate 

the road bed of Power Creek Road so it is not subjected to flooding, this would constrict flows and 

possibly make downstream spawning areas more susceptible to scour. 

Aquatic Vegetation 

While increased atmospheric carbon dioxide levels and a longer growing season are generally expected to 

increase plant growth in Alaska (Haufler et al. 2010), site specific factors and individual species 

responses make it difficult to predict the overall effect in wetland communities (Poff et al. 2002). Eyak 

Lake already has large areas covered by aquatic plants, including various species of Potamogeton and the 

non-native Elodea canadensis. If these species respond positively to climate changes, there may be 

adverse effects to fish habitat. 

One potential effect is that increased amounts of vegetation could lead to greater biological oxygen 

demand under the winter ice when the plants die and decay. In areas where there are insufficient 

streamflows entering the lake, localized anoxic zones could develop. This risk could be reduced if warmer 

temperatures keep the lake surface ice-free for a greater part of the winter. 

Eyak Lake Watershed Management Recommendations 

The most important part of these climate change analyses should be determining what can and cannot be 

done, or at least what should or should not be done. 

Most of the current problems, stressors, and potential risks for the Eyak Lake watershed are outside of 

National Forest land or are issues not managed by the Forest Service. There are, however, some actions 

that can be taken either unilaterally by the Forest Service or in conjunction with cooperating agencies and 

organizations. For the values identified for the Eyak Lake watershed, protecting the salmon stocks and 

adopting measures to mitigate the predicted increase in flooding are the primary concerns. 

Forest Service Management 

The current Forest Plan manages most of the upper watershed as a “primitive” area, while other areas 

have restrictive covenants that were established when the land was purchased from a local Native Alaskan 

corporation. The area is not available for timber harvest, and while mineral development is conditional, 

there are no active claims and no known mineral resources. There are no Forest Service roads. No offroad 

vehicle use is permitted. Thus, management actions are limited, and with the relatively pristine state 

of the National Forest land, there may not be much that can be done to improve conditions in preparation 

for climate change. 

There have been suggestions that large woody debris (LWD) could be added to streams to moderate flows 

or provide refugia for juvenile fish, which could buffer the effects of predicted high flows or floods. This 

can be useful where natural sources of LWD have been removed or in highly disturbed areas (Bair et al. 

2002). However, Bakke (2008) points out that areas affected by climate change are likely to be unstable 

and any structures or stream engineering will have to be carefully designed to accommodate change. 

Redundant structures are recommended in anticipation that many structures may fail or may not have the 

intended effect. 



Bakke (2008) advises, “Passive restoration techniques, such as establishment of wider riparian buffers, 

may be a more sustainable alternative in light of increased geomorphic instability caused by global 

warming.” This may well be the case in the Power Creek delta, where sediment from landslides and 

exposed glacial moraines will be deposited and where channels can be expected to fill and shift 

frequently. 

Thus, it may be best, and less costly, not to alter naturally functioning channels. Maintaining the current 

floodplain connectivity may do the most to protect fish habitat from floods and scouring of redds. 

Keeping the upland vegetation and slide-prone slopes undisturbed should be the key methods for 

minimizing runoff, landslides, and transport of material to the streams. 

If development projects are proposed, managers would obviously need to be aware of the increased 

potential for avalanches, landslides, and flooding in project areas. There will also be a need for more 

appropriate road construction standards, such as more frequent cross drainage, larger culvert size, and 

more consideration of slope stability. 

Cooperative Efforts 

Flooding 

The most likely adverse effect of climate change will be the increased frequency of floods, which will 

affect residences, small businesses, and other development along Eyak River, as well as areas around the 

lake. Flood mitigation measures will require cooperative efforts among government agencies, private 

landowners, and Native corporations. Assuming that the uplands will be managed properly, the question 

becomes what other actions can be taken to prevent flooding or to mitigate the effects. 

One project that has been proposed over the past 25 years is to build a dike separating Eyak River and the 

glacial Scott River. As mentioned above, the Scott River can deposit sediment in lower Eyak River, 

reducing the Eyak channel’s drainage capacity. The project has never been implemented, due to the high 

construction and maintenance costs. Project investigators for the U.S. Army Corps of Engineers stated in 

2000 that the dike would cost $5 million to $8 million, and given their hydrologic data at the time, the 

value of the property and houses that might be flooded was only $2 million (Hodges 2000). 

Given the predictions of more frequent flooding, possible higher flood levels, and the increased 

development and property value in the area since that time, it would be reasonable to study the situation 

and cost/benefit analysis once again. One specific action that is needed is to develop a “water budget” for 

the watershed, as proposed by Rothwell and Bidlack (2011). At the present time, there is no way to 

correlate streamflows, precipitation, etc., with lake and river levels and, in turn, flood levels. Once a water 

budget is developed, predicted increases in precipitation and other climate change information can also be 

incorporated for determining flood risks in the future. 

One other flood issue is the potential water pollution from fuel and other substances stored in flood-prone 

areas. Almost all of the residences rely on fuel oil for heating, and the tanks are susceptible to damage or 

inundation. Through its Million Dollar Eyak Lake program, the Copper River Watershed Project is 

looking into ways to get homeowners to elevate fuel tanks above flood levels and to adequately secure 

tanks so they are not washed away. Public education and possible grant opportunities for implementation 

are being considered. Many landowners have already begun raising their tanks and houses, as well. 



Eyak Lake Area Meriting Special Attention (AMSA) Cooperative Management Plan 

The Copper River Watershed Project is working on an update of the Eyak Lake AMSA plan (Professional 

Fishery Consultants 1985) that assesses the condition of Eyak Lake, which was designated as an “area 

meriting special attention.” The ADFG, Prince William Sound Science Center, City of Cordova, Native 

Village of Eyak, Ecotrust, USDA Forest Service, and others have worked together identifying resource 

issues, community concerns, monitoring needs, and possible projects for restoration or to improve 

recreational uses. 

Some of the issues identified include non-point source pollutants, effects of the Power Creek Road and its 

culverts on the lake and spawning areas, pollution from the flooding of developed areas along Eyak River, 

and relocating a boat ramp. While these issues do not directly relate to climate change, maintaining the 

health of the watershed and its fish and wildlife species, is perhaps the best way to mitigate potential 

effects in a system that is generally functioning in a natural condition. 

Fisheries Management 

The Forest Service has no direct management authority over fish populations but sport and subsistence 

fishers are important users of National Forest lands in the Cordova area. The nutrients that spawning 

salmon bring to the watershed are also an important part of the ecosystem, not only for predators such as 

bears and eagles, but for future generations of salmon as well (Lang et al. 2006). Thus, it is important to 

have sufficient numbers of salmon returning to streams in National Forests and for the Forest Service to 

provide input and assistance where possible. 

Just recently, the CRWP and the Prince William Sound/Copper River Marketing Association (a 

commercial fishing group) recently started an outreach to see if there is interest in developing a 

sustainability plan for the Copper River and Prince William Sound fisheries. The announcement stated, 

“Our goal is to bring together information resources on fisheries, management and habitat; identify data 

gaps and information needs; and identify indicators for tracking sustainability of the fisheries over time.” 

(CRWP and PWSCRMA 2011.) 

This appears to be a good cooperative opportunity for agencies, organizations, commercial interests, 

Native groups, and others to provide input for the managers at the ADFG. One example would be the 

management of the coho salmon fishery. Currently, coho salmon in the Copper River Delta and adjacent 

systems are managed as a single stock based on aerial observations of index streams. There are no set 

escapement goals for individual streams; rather, the management biologists work to meet an overall total. 

In practice, the desired range of the combined counts has been met consistently (Botz et al. 2010). 

Hilborn et al. (2003) and Bryant (2009), however, suggest that genetic stocks may occur on a much 

smaller level, either among or within stream systems. Ruff et al. (2011) identified distinct genetic stocks 

associated with different spawning behaviors within a single system. Thus, to maintain the ability of a 

species to adapt to change, especially in their behaviors, diverse stocks need to be preserved. Bryant 

(2009) concludes that in view of the potential disruptive effects of climate change, future harvests should 

be conservative to ensure that all stocks have sufficient escapement. 

Given the satisfactory overall counts, the management strategy appears to be working well under the 

present conditions. However, in order to conserve all of the stocks, interested parties should collaborate 

on ways to monitor escapement in the numerous smaller systems. The Forest Service and other partners 

could take an active role and provide additional personnel to obtain this information and ensure that the 

current management is effective. 



Watershed Restoration 

Restoring damaged watersheds to improve their natural function is widely seen as the key to increasing 

resiliency to the effects of climate change (Furniss et al. 2010, Haufler et al. 2010). The Eyak Lake 

watershed has not been severely damaged, but there have been restoration opportunities, and some still 

exist. 

The CRWP has taken the lead in implementing restoration projects through their FishWatch and Million 

Dollar Eyak Lake programs. The Forest Service, ADF&G, Native Village of Eyak, and others have 

worked with CRWP to identify and prioritize projects. Some of the specific projects have included the 

following. 

• Replaced three undersized failing culverts with an arch culvert that restores passage to upstream 

fish habitat and downstream transport of spawning gravels to sockeye salmon spawning areas in 

the lake. 

• Installed a Stormceptor oil and grit separator to remove sediment and hydrocarbons from street 

runoff and an urban stream that flow into the lake. 

• Removed an artificial spit and abandoned floatplane dock that adversely affected sockeye salmon 

spawning habitat in the lake. 

• Revegetated disturbed shorelines where roads border the lake. Vegetation will reduce erosion, 

trap sediment runoff from the roads, and provide shade and cover to improve fish habitat. 

• Worked with the City of Cordova to address snowplowing and dumping practices to help keep 

sand, salt, and hydrocarbons from entering the lake. 

Thus, many of the existing problems have been addressed. There are still some culverts that prevent fish 

passage, but the loss of habitat is relatively small, and replacement costs would be high. The CRWP, in 

partnership with Ecotrust, ADFG, the US Fish and Wildlife Service, and the Alaska Department of 

Transportation, has also developed a culvert replacement prioritization protocol that has been used in the 

Eyak watershed and surrounding areas (CRWP 2011). The highest priority sites are outside of the 

watershed. 

Eyak Lake Watershed Summary 

The Eyak Lake watershed was chosen because it is typical of coastal Alaska and because its climate 

change issues would be similar for most rainforest watersheds in southcentral and southeast Alaska. 

Higher precipitation, melting glaciers, and more frequent rain-on-snow events increase the possibility of 

floods, erosion, increased sediment transport, and changes to channels in depositional areas. All of these 

increase the risks to infrastructure and fish habitat. 

As discussed by Rothwell and Bidlack (2011) there are many data gaps that hinder the development of a 

water budget for Eyak Lake, therefore, it is difficult to quantify flows and their effects. There are also no 

models that can predict and quantify snow avalanches and how they affect the landscape. However, a 

general look at the issues and values allows land managers to identify possible mitigation actions, or 

things to leave as is – in this case the existing flows and habitats that appear to be functioning well. 

Maintaining the habitat and the diverse genetic stocks may be all that mangers can do to buffer the effects 

of climate change. 

This brief study also shows the value of an active, concerned community. NGO’s like the CRWP have 

taken an extensive role in identifying and implementing restoration projects. The Prince William Sound 

Science Center, Ecotrust, and others are conducting studies that will provide baseline data for future 



assessments. Again, the watershed is generally functioning properly in a natural condition, as is 

evidenced by its abundant fisheries resources. However, the watershed needs to be managed well, 

maintained, and monitored to continue its productivity. The local community and user groups that derive 

the benefits of the resources are probably the best stewards. 

Resurrection Creek Watershed 

Figure 5. The Resurrection Creek watershed association. The town of Hope and the areas along the coast lie outside 

of the Resurrection Creek watershed. 

Area Description 

The Resurrection Creek watershed was added to this assessment to examine the issues and conditions on 

the western side of the Chugach National Forest. Although the watershed is coastal in the sense that it 

drains directly to saltwater, the mountains and prevailing storm patterns reduce the precipitation, giving 

the watershed a drier climate. Potkin (1997) describes the Kenai Peninsula as a transitional area between 

the coastal rainforest and the inland boreal forests. Climate change predictions, however, call for 

increasing temperatures, particularly in winter, and increases in precipitation. 

The Resurrection Creek watershed is a U-shaped valley with steep slopes, a low- to moderate-gradient 

valley floor, and a dendritic stream drainage pattern. The tributary streams are generally steep and form 

alluvial areas as they reach the floor. 

This watershed is a popular recreation area and has five species of Pacific salmon; it also has a history of 

hydraulic mining, forest insect infestation, and occasional wildfires. Mining has been the most disruptive. 

The natural tributary channels have been diverted to power hydraulic cannons (Kalli and Blanchet 2001), 

while the main creek has been diverted from one side of the valley to the other for easier access to the 

alluvial deposits. A one-mile section of the upper creek has had extensive restoration work but the lower 

creek still has substantial problems. 



The town of Hope (population 182) lies near the mouth of the creek; most of the residences and 

development are outside of the watershed. The town is supported mainly by tourism. The historic 

buildings and a modest pink salmon recreational fishery are the main attractions. Commercial miners 

have claims to old tailings piles and alluvial material in the lower floodplain but activity has been 

sporadic. There is no industrial, agricultural, or other large-scale use of water. The mining activities 

occur at a level that does not require large diversions of water. The water supply for the town comes from 

private wells. 

Watershed Values 

• Recreational fishing, primarily for pink salmon. 

• Five species of Pacific salmon (peak counts): chinook (600), chum (892), coho (900), pink 

(40,000), and sockeye (37). 

• Resurrection Pass Trail in the main valley: 19 miles of trail and three Forest Service recreation 

cabins. Popular for summer and winter recreation including hiking, mountain biking, 

snowmachining, skiing, and snowshoeing. 

• Recreational gold dredging and gold panning. 

• Limited commercial mining operations on floodplain. 

• Limited residential structures and tourist oriented businesses within the watershed, to which the 

town of Hope is immediately adjacent. 

Data Available, Data Needs 

• Air temperature and precipitation collected 1979-1995. Some data are missing. Permanent station 

at Moose Pass, 25 miles south. 

• United States Geologic Survey Stream Gauge 1967-1986. 

• SNAP program conducted by the UAF has predictions for temperatures, precipitation, and 

freeze/thaw dates at a 2km scale. This was calculated with PRISM and five climate models. Online 

maps and bar graphs are available for Alaska communities. Raw data is available for use with 

GIS. 

• Global Land Data Assimilation System (NASA 2011) has soil moisture, evapotranspiration 

estimates using VIC for 1979 to present, but no future estimates yet. Different models show 

conflicting results for amounts and increases in evapotranspiration rates but two of three show 

increases for 1979-1991 compared to 1992-2010. 

• Limited data for the stream restoration work in upper Resurrection Creek are available. 

• Additional data are needed for total fish habitat and for miles of stream still disconnected from 

the floodplain by tailings piles and channelization. 

Resurrection Creek Sensitivity and Stressors 

As with many mountainous areas, there are steep, unvegetated slopes at the higher elevations, which are 

prone to snow avalanches and landslides. Avalanches occur in most of the tributary streams during winter 

and spring, providing a source of colluvial sediment along the streams (Kalli and Blanchet 2001). At 

lower elevations, the thick vegetation, relatively low precipitation, and low precipitation intensity and 

duration reduce flashy flows, stream bank erosion, and surface erosion (Kalli and Blanchet 2001). 

Human derived stressors are mostly confined to the valley floor where mining has severely altered 

channels and flow patterns. Mining has affected about 2,560 acres of the floodplain along the main stem, 

as well as patches along a one-half mile stretch at the mouth of Palmer Creek. The mining-caused 

problems that may be exacerbated by climate change include: 



• Tailings piles deposited along the creek have confined the channel and caused downcutting. 

Greater flow velocities, scouring, and erosion may occur with predicted increases in precipitation 

and extreme events. Salmon redds may be subject to scouring. 

• The creek channel has been moved and straightened, increasing the gradient and water velocity. 

Again, scouring and erosion are likely to increase with precipitation and extreme events. 

• Mining has removed the trees in the riparian areas. This has resulted in the loss of future LWD 

that would add roughness to the channel and moderate water velocities. The loss or pool-forming 

LWD reduces fish habitat. 

• Mining activity has reduced the fine-grained sediment and organic material from the floodplain, 

so re-establishment of the riparian vegetation has been minimal. Without healthy vegetation, the 

streambanks are more sensitive to erosion from high flows during extreme events. 

The topography and current climate conditions, however, may reduce the sensitivity of the watershed to 

climate change. Even with the predicted increases in temperature and precipitation, the watershed will 

still remain relatively cold and dry. In addition, some current and proposed restoration work could lessen 

the sensitivity. In brief: 

• Cold winter temperatures (even at sea level), high mean elevation, low precipitation, could all 

reduce sensitivity to snowline increase and rain-on-snow events. 

• Continued cold winter temperatures at high elevations should result in fewer freeze/thaw cycles 

and instances of wet heavy snow falling on dry snow layers. Avalanche danger and its sediment 

transport may not be sensitive to changes. 

• Short duration, high intensity storms are relatively rare, and the flow response from such events is 

limited by high initial infiltration. This reduces sensitivity to flooding and high flows. 

• A restoration project along one mile of stream reconnected floodplain, created meanders to 

reduce gradient, added LWD and secondary channels. Several miles of unrestored channel 

remain. 

• Current low stream temperatures, minimal lake area, make water temperatures less sensitive to 

warming. 

• Low acreage of glaciers/permanent ice field reduces sensitivity to the effects of increased glacial 

melting – higher flows, moraine transport. 

• Low road density and minor current mining operations do not contribute significant amounts of 

sediment to the streams. 

One non-aquatic stressor that may have already been worsened by climate change is increased timber 

mortality due to the spruce bark beetle. Warmer winters have been cited as one reason for increases in the 

beetle population and infestation of the stands. Continued warming trends could lead to further increases 

in the beetle population and greater tree mortality. 

With high numbers of dead trees, the watershed is expected to be more vulnerable to fire, although the 

extent of risk is in question. Fire and the resulting loss of vegetation could lead to greater erosion of the 

hillslopes. More in-depth analysis of the fire potential is needed, but the general outlook is that the risk of 

fire will increase as described here: 

• The spruce bark beetle infests about 11% of the watershed, resulting in high levels of dead trees 

and fuel loading. Predicted warmer summer temperatures with only small increases in 

precipitation may increase fuel drying and fire hazard. 

• Increased temperatures, growing season, and precipitation could increase grass and shrub growth, 

increasing fuel load (Haufler et al. 2010). 



• One source suggests the Resurrection Creek watershed is insensitive to wildfire. Historically, 

fires are infrequent and of low intensity given the moderate levels of precipitation compared to 

the western Kenai Peninsula. The general north-facing slope aspect reduces sensitivity (Kalli and 

Blanchet 2001). 

• Another source says that Hope and nearby communities are at greater risk. An interagency plan 

states the Hope/Sunrise area is at a high risk – on a scale of low, moderate, high, and extreme 

(Kenai Peninsula Borough 2004). Part of this rating may be due to the poor road access and 

availability of personnel and equipment. 

Trends 

The population in the Hope area increased from 137 in 2000 to 182 in 2010 but there is a large margin of 

error (U.S. Census Bureau 2011). There has also been an increase of 31 housing units, some of which 

may be cabins or other development targeted for tourism. Most of the development is in areas adjacent to 

the Resurrection Creek watershed, but this still increases exposure to wildfire along the wildland urban 

interface. 

The increased development suggests that there is more interest in the area and probably more use of the 

recreational opportunities within the watershed. No figures are available for future recreational use 

specifically for the Resurrection Creek area, but recreation use and tourism are projected to increase 

throughout the Kenai Peninsula (USDA Forest Service 2002). 

The situation with commercial gold mining is unclear at the present time. There have been discussions 

between the Chugach National Forest and the mining interests regarding stream restoration work the 

Forest Service would like to implement in the lower stretches of the creek. However, this and other 

information on future mining plans are not available. 

A watershed restoration project along a one-mile stretch of Resurrection Creek should provide significant 

benefits to the hydrology of the system. The biological benefits will arrive more slowly, but are expected 

nonetheless. Fish populations should increase with habitat improvements (Martin et al. 2010), particularly 

for coho salmon. Because only two brood years of coho salmon have returned since the completion of the 

project, not enough time has passed to detect any trends. 

The riparian vegetation that was planted at the project site should be established by now but it will still 

take several more years for the shrub species to reach maturity. Sitka alder (Alnus sitchensis) should also 

be regenerating naturally. Conifers will require many decades to reach a size large enough for meaningful 

input into the stream as large woody debris (Farr and Harris 1979). 

Exposure/Risks 

The predicted changes call for increases in precipitation and air temperatures, as well as a reduction in the 

number of days below freezing, summarized in Table 1. There are conflicting results for changes in 

evapotranspiration from 1979 to 2010 (NASA 2011), but it appears that rates in the Kenai Peninsula area 

have been increasing (Haufler et al. 2010). 

Fire Hazard Risk 

One of the main concerns on the Kenai Peninsula has been the risk of fire, because many of the smaller 

towns such as Hope are within or adjacent to forests. The towns’ isolation, relative lack of firefighting 

personnel, and lack of equipment make these communities especially vulnerable. In addition, fuel loads 

are high, due to the number of spruce killed by infestations of the spruce bark beetle. 



Historically, wildfires have been infrequent and of low intensity in the Resurrection Creek watershed, but 

predicted increased temperatures and higher evapotranspiration rates may dry fuels and increase the 

number of hazardous fire days on the peninsula as a whole (Haufler et al. 2010). Earlier snowmelt dates 

could also extend the period during which grasses and other dead vegetation can dry and provide 

flammable material, before the spring green-up, thus extending the fire season (Ecology and Environment 

Inc. 2006). 

Hydrologic/Geomorphic Risks 

The predicted annual increase in precipitation is relatively small at 2 to 3 inches, so unlike rainier areas of 

the Chugach National Forest, flooding may not be seen as a great concern. High initial infiltration rates 

also reduce the risk (Kalli and Blanchet 2001). An increased risk of floods from rain-on-snow events may 

not be likely. Hamlet and Lettenmaier (2007) state that cold systems where snow processes dominate the 

hydrologic cycle may be less prone to flooding. Even though winter temperatures are expected to increase 

by 4 °C, mean temperatures at sea level are still predicted to be below freezing. The usual rain-on-snow 

events in the fall and spring may occur, but not throughout the winter, as they would in a warmer area. 

The town of Hope and the infrastructure in the Resurrection Creek valley are unlikely to be affected by 

floods. Most of the town straddles a low ridge between two watersheds. The town’s buildings are set on 

higher areas away from where the creek enters the ocean. The buildings, roads, and the airstrip in the 

valley are also on relatively high ground. There are only 20 developed parcels within the 100-yr 

floodplain (Kenai Peninsula Borough 2011). Despite frequent extreme weather in recent years, which has 

caused flooding in other areas of the Kenai Peninsula, no flood damage was reported for the Hope area 

(Kenai Peninsula Borough 2011). 

A later freeze date and earlier spring melt would change flow timing, however. The current peak 

discharge is in mid-June and would be expected to occur earlier. Reduced flows from July to September 

could be partially offset by increases in precipitation ranging roughly from one-half to one inch of rain 

per month. Also, given the high mean elevation of the watershed and the predicted increase in winter 

precipitation, there could be an increased snowpack at the higher elevations that would last longer into the 

summer. 

In warmer, rainier areas, erosion and sediment transport are expected to increase because of higher 

precipitation, rain on snow events, increased freeze/thaw cycles, avalanches, and exposed glacial 

moraines. The Resurrection Creek watershed, however, should be less exposed to these factors because of 

the low predicted increases in precipitation, low winter temperatures even with warming, and limited area 

of glaciers and icefields. Thus, the risk of increased filling and shifting of channels is not expected to be 

much higher than existing levels. 

Biological Risks 

Although the Kenai Peninsula is a relatively dry area for Alaska, low flows are not expected to be a major 

concern for fish. As discussed earlier, the changes in the hydrograph may be offset by increases in 

summer precipitation and an increased snowpack at high elevations. The lowest flows are in the winter, 

and given the warmer temperatures, precipitation falling as rain in early winter could increase flows then. 

The risk is also lessened by the fact that most fish habitat is in the low gradient channels near the valley 

floor, rather than small headwater streams. These lower elevation streams drain larger areas and are less 

likely to dry up. 

Higher precipitation is not likely to increase the risk of salmon redds and juveniles being scoured by high 

flows, given the moderate increases. This risk is probably more dependent on other factors, such as the 



channelization of the stream from mining activity. The recent restoration work that lowered the stream 

gradient and restored floodplain connectivity should provide some help to reduce the risk of redd scour, 

however, there has been no monitoring of this yet. The use of lower-velocity side channels as spawning 

areas by coho, chum, and pink salmon also reduces the risk of redd scour. 

The issues related to warmer water temperatures that were discussed in the Eyak Lake watershed section 

also apply here. Resurrection Creek and its tributaries are currently cold enough that increased water 

temperatures will not be beyond optimal temperatures. However, unknown problems may be caused by 

faster egg development, increased metabolic rates, and the desynchronization of life-stage timing with 

other biological and physical conditions. 

The biggest risk for salmon is that their populations are small, except for pink salmon. A disastrous event 

or adverse conditions for several years could extirpate the less abundant species, particularly the sockeye 

salmon, which generally do not use systems without large lakes. A large fire could be such an event if it 

removes vegetation and leads to significant erosion, sedimentation, or channel changes. As discussed, the 

fish habitat in the system has been highly disrupted already. More restoration work is planned, but until 

the salmon species become more established, they will remain susceptible. 

Resurrection Creek Management Recommendations 

There has already been extensive planning for the Kenai Peninsula area, including the Resurrection Creek 

watershed. The Kenai Peninsula Borough, in cooperation with other partners, has developed the All- 

Hazard Mitigation Plan that includes strategies for addressing eight hazards, including wildfire, floods, 

weather, and avalanches. There are detailed action plans, mitigation measures, hazardous site evaluations, 

and ideas for future actions and cooperative efforts. Thus, there is little need for land managers to reinvent 

the wheel; there might simply be a need to continue the ongoing work while keeping the implications of 

climate change in mind. 

A primary concern for land managers is public safety, and there is an immediate risk of wildfire near the 

town of Hope. The wildfire section of the Mitigation Plan includes specific goals for fuel reduction, 

controlled burns, fire breaks, and public education. The Chugach National Forest has completed its first 

five-year action program under this plan and is now working on strategies for the next five years. Actions 

in the Resurrection Creek watershed have included controlled burns, seeding areas with birch (in place of 

spruce, which is susceptible to beetle kill), and working with private landowners to make structures less 

vulnerable to fires. 

The Mitigation Plan also addresses the danger of snow avalanches, which increases with variable 

temperatures creating layers of wet and dry snow. Although mean winter temperatures are expected to 

remain below freezing in the Resurrection Creek watershed, the high degree of winter recreational use 

makes it an issue to be dealt with. The Forest Service currently operates the Chugach National Forest 

Avalanche Information Center, which provides recreationists with current snow conditions. The Forest is 

also hiring a meteorological technician to help with this program. This is another example of how existing 

programs can address future risks. 

The other main action that can be taken in the watershed is to continue with the stream restoration 

program. Although there are some conflicts between existing claims and the areas to be restored, 

returning the channels to a more natural condition will provide the best long-term protection from floods 

and for fish habitat. 

The costs are significant. The previous restoration project cost about $700,000 per mile, and simply 

removing the tailings piles to establish floodplain connectivity might cost $300,000 per mile. However, 



reducing the risk of redd scour and creating additional high-quality fish habitat would help maintain and 

improve the small salmon populations that are becoming re-established, and would help maintain the 

recreational fisheries that draw people to Hope. Smaller scale work, such as adding more LWD to offchannel 

rearing areas and secondary channels could provide some benefit since the past mining operations 

removed the natural LWD and the riparian trees (Martin et al. 2010). 

Resurrection Creek Watershed Summary 

The Resurrection Creek watershed was chosen because it is more typical of the drier, colder Kenai 

Peninsula climate. Given these conditions, and the relative lack of infrastructure in the watershed, the 

predicted increases in temperature and the relatively small increase in precipitation are not expected to 

have as great an effect as in other areas. 

Management direction is also made simpler because of the existing plans for addressing wildfire, snow 

avalanches, and watershed restoration. The actions that are already outlined in these plans appear to be the 

same steps that should be taken to mitigate for climate change. Managers should review the plans in light 

of the predicted changes – accounting for higher flows when reconstructing stream channels, for example 

– but the basic direction and schedule of work appears to be what is needed. 

DISCUSSION AND GENERAL GUIDELINES FOR MANAGERS 

Given that most of the watersheds on the Chugach National Forest are essentially unaltered and are 

functioning naturally, this assessment was limited to two specific watersheds where there has been some 

development and where at least some hydrologic and climate data were available. The intent, then, 

became not to identify which watersheds in the Forest were the most vulnerable, but rather to look at 

these two watersheds and identify specific vulnerabilities and possible mitigation. 

This assessment was limited in some ways, not because of the lack of predicted climate change data, but 

because of the problems associated with drawing specific conclusions from the data. As mentioned 

earlier, without data on lake and river levels in the Eyak Lake watershed, it was not possible to determine 

the specific risk of floods, although existing conditions and climate predictions point to greater risks. The 

predictions of the increased frequency of extreme events also make it difficult to determine risk. 

Looking at the effects of climate change from a general viewpoint can be valuable, despite the 

uncertainties. By examining specific watersheds and issues, land managers can determine actual on-theground 

actions that can be taken to help mitigate the effects of climate change even if the specific degree 

of risk is not known. The following sections discuss this approach, which was used for this assessment, 

along with some shortcomings and lingering questions. 

Climate Change Data Acquisition and Analysis 

For Alaska, there is a considerable amount of predictive climate change data available online; the main 

question is how to analyze and apply it to specific areas. It is relatively easy for a competent GIS user to 

manipulate mean temperature and precipitation values, but determining how these changes might affect 

flows and salmon habitat requires many additional levels of information. It may be easy to get caught up 

in the GIS data, while losing sight of what it actually means on the ground. Thus, before delving too 

deeply into data analysis, managers need to determine what they need and how they can use these results 

in a practical manner. 

One other type of analysis that was attempted was NetWeaver, a knowledge-based decision support 

system using fuzzy logic. It can be used when data are not complete, and expert opinion or other means 



can be used for the gaps. The system tests whether a statement is true (e.g., coho salmon habitat is 

suitable), based on a number of dependent data inputs. The validity of this method is discussed by 

Reynolds (2001), and the method was used for the Aquatic and Riparian Effectiveness Monitoring Plan 

for the Northwest Forest Plan (Reeves et al. 2003). Some of the key benefits are that it forces managers to 

analyze issues in a clear, rational method, and that the links between factors and conditions, the causes 

and effects, can be clearly diagrammed. 

Using this tool may seem easy in concept, but can become exceedingly complex, even for simple 

biological questions. Summer coho salmon habitat, for example, will depend on water temperature, cover, 

food, water velocity, etc., all of which may depend on multiple subfactors. Temperature could depend on 

riparian vegetation, groundwater input, stream width, and so forth. At each step, one has to determine 

how much to weigh each factor in relation to other factors and how to evaluate each factor. The degree of 

uncertainty seems to accumulate with every estimation, opinion, or assumption. Thus, one can spend a lot 

of time working out the details of this analysis method and never reach a conclusion with which one feels 

comfortable. This was my experience. 

If one wants to use NetWeaver or some similar method, it would be best to have experienced users to 

point out the limitations, particularly as to the level of investigation. Deriving broadscale conclusions for 

the Northwest Forest Plan is probably more appropriate than trying to analyze conditions in a small 

watershed, where you would want more detailed answers. Also, since many inputs may require expert 

opinion, it would be best to have a number of qualified people to present their views for each topic 

(Reynolds 2001), not just a single person. Even though NetWeaver may reduce the need for some data, it 

still requires a good deal of intellectual input and effort to get a satisfactory product. 

So, as far as analysis is concerned, a general idea of the types and magnitudes of climate change – which 

could be readily available from the internet – may be enough to get started. The key first step might not 

be to obtain specific numbers, but to analyze how those changes might generally affect the resource 

values in a given area. After that, one can determine if there is anything that can be done about the 

problem, and how much more specific data is needed for project implementation. Again, local groups 

with existing plans, such as the Kenai Peninsula Borough’s All Hazard Mitigation Plan or the Copper 

River Watershed Project’s Million Dollar Eyak Lake program can provide direction or ready-made 

solutions. 

Direction for the Future 

Once managers have looked at the resource values and how they might be affected by climate change, 

there is the need to implement the mitigation proposals. Certainly, there is a laundry list of tasks that can 

be applied to almost all areas and that should be implemented as a normal course of work. Some 

examples include: 

• Replacing “red” culverts that are inadequate for fish passage or flows. Replacement culvert sizes 

will need to be adjusted for predicted flows under climate change scenarios and extreme events. 

Utilize existing culvert prioritization protocols. 

• Maintaining roads at least to current standards. In the long term, standards should be reviewed in 

light of predicted climate changes, such as requiring more frequent drainage structures for areas 

with increased precipitation. 

• Examining infrastructure in riparian or other areas that may be subject to floods or snow 

avalanches, in regard to public safety. 

• Restoring existing damaged riparian areas, particularly in regard to floodplain connectivity in 

areas susceptible to floods from rain-on-snow or extreme events. 



• Restoring riparian vegetation to maintain cooler water temperatures. 

There are many other tasks, but the underlying theme is that fixing existing problems can go a long way 

toward mitigating climate change effects. However, to be most effective, the current engineering or 

biological standards should also be reviewed and adjusted in view of the predicted changes. 

One other area where managers can be effective is through reviews of their Forest Plans. As an example, 

one of the biggest issues in the past, for the Chugach National Forest, has been winter recreational use. 

The Plan is up for review and managers might consider the possibility of reduced recreational 

opportunities from shorter winters and higher snowlines. There may be a need to open new winter 

recreation areas, rewrite management prescriptions for existing uses, or improve access to higher 

elevation areas. 

If managers wish to be proactive about climate change, a committee could look at each component of the 

Forest Plan to see how it might be affected by predicted changes. Some areas may need little or no 

adjustment, and monitoring baseline conditions might be sufficient. The Chugach plan has a Monitoring 

and Evaluation Strategy, which would be the best place to establish a climate-change monitoring design. 

In any case, the Forest Plan is one place where managers can establish policy and show commitment 

toward addressing climate change. 

Biological Issues 

The biggest lingering question is how species, particularly the highly valued salmon species, will respond 

to climate changes. Unlike areas in the lower 48 states, the freshwater changes in coastal Alaska are less 

likely to have direct lethal effects to salmonids, but life-cycle timing and changes to food source species 

could occur. Although Haufler et al. (2010) state that a risk assessment needs to be made for Alaska 

salmon, knowing how salmon will respond to the predicted changes and trying to assign risk appear to be 

difficult tasks at this point. 

As mentioned in the Eyak Lake watershed discussion, part of the salmon response will depend on the 

response of other organisms, especially whether the life cycles of prey species change in synchrony with 

newly emerged fry. This is not presently known. The other part of this situation is how well a species 

itself can adapt to changing conditions. If, for example, warmer temperatures cause fry to hatch too early 

in the spring, does the species have the innate capacity to adjust its spawning to a time later in the fall to 

compensate? 

It would appear that this capacity does exist for some salmon species that have a diverse life history. One 

example is a groundwater-fed spawning channel near Cordova used by coho salmon. The adults spawn 

over a wide period of time, from October well into December, with fry emerging from May to mid-July 

(unpublished Forest Service data). If warmer groundwater temperatures cause faster development, but the 

optimal hatching time continues to be in June, the progeny of late-December spawners could still sustain 

the run and adapt over time. Such diversity may make these species more resilient to change, assuming 

that food chains or other conditions are not totally disrupted by climate change. 

Another part of this question is how well species will survive climate changes, given the highly variable 

weather conditions that already exist in an area like Cordova. From 1949 to 2004, the mean annual 

temperature at the Cordova airport has been 39.1 °F, but the extreme annual temperatures have ranged 

from 34.3 to 41.4 °F. Annual precipitation has averaged 96 inches, but has ranged from 54 to 139 inches. 

There is no certainty that species will be able to cope with extended years of the predicted higher 

temperatures and precipitation, but the species of the area have survived conditions similar to what is 



predicted at least on an occasional basis. It is certainly speculative, but the various species may already 

have the genetic capacity to persist considering their past experience. 

Along these lines, Bryant (2009) points out that the future Alaska climate may become more like British 

Columbia, where the same salmonid species exist, or have existed, in abundance. In time, with the help of 

straying or through selection, these species could be expected to exist or even flourish in Alaska under the 

changed climate conditions. The only difference, Bryant notes, is that the evolution will need to occur 

over a period of decades rather than hundreds or thousands of years. 

Thus, the key to maintaining species of all sorts may simply be through the conservation of diverse 

habitats and genetic stocks (Hilborn et al. 2003, Bryant 2009). Although many habitats in southeast 

Alaska have been damaged by timber harvest or other management, Bryant (2009) states that there are 

still numerous unaltered watersheds that can buffer the effects of climate change. Timely restoration work 

in the altered areas can help to save stocks that are in danger. 

This is not to say that there will not be adverse effects while the populations are adjusting to the new 

conditions and stresses. In regard to salmon, Bryant (2009) stresses the potential need for cooperation 

among all users groups to manage conservatively and reduce harvests, even if population stresses are not 

readily apparent. Since we cannot determine the genetic composition of fish in every stream and habitat 

niche, the management strategy should be to ensure that all existing stocks, based on locations and run 

timing, have sufficient returns. 

Current Research, Monitoring 

As discussed in the previous section, much of the uncertainty about risk is due to a lack of understanding 

about the biological processes and how species will respond. In addition, some basic parameters, such as 

groundwater flows and temperatures, have not been studied. Researchers from the Pacific Northwest 

Research Station and various universities are attempting to fill these knowledge gaps with a number of 

studies on the Copper River Delta area, including some monitoring sites in the Eyak Lake watershed. 

One study examines the life-history diversity of populations of coho and sockeye salmon in streams with 

different seasonal thermal regimes. These differences may be related to location, groundwater input, 

glacial melt, and surface water input. Using scales and otoliths from adult fish returning to spawn at these 

sites, researchers will determine a number of life history parameters including size at emergence, number 

of years spent in freshwater, and size at ocean entry. 

If differences are correlated with varying temperature regimes, researchers may be able to predict what 

might occur from the changes associated with climate change. For example, warmer winter air 

temperatures may lead to increased amounts of surface water input in a system, as precipitation occurs as 

rain rather than snow. The temperature change may then affect the incubating eggs and their rate of 

maturation. 

Another ongoing research project is a study of aquatic invertebrates in ponds with different temperature 

regimes – some located in the relatively warmer west Copper River Delta and others in the colder east 

delta. Again, location is used as a surrogate for the temperature changes that are predicted over time. 

Differences in larval development, emergence timing, and possibly the annual number of generations of 

some species, could have a significant effect on predators. This could be especially true for avian species 

whose migratory patterns may be based on daily photoperiods rather than temperature. 

There are a number of other research questions that should be asked for Alaskan areas, especially in 

regard to the ability of species such as salmon to adapt to changed conditions. Also, in rural or remote 



areas, there is likely a need to collect simple baseline data, such as groundwater temperatures, that can be 

monitored over time to detect or verify predicted changes. 

Most of all, managers need to determine how the information is going to help make on-the-ground 

decisions. Certainly, we would like to know that if fry develop faster and emerge earlier, their food 

resources will also develop faster and will be available. But we need to be thinking about how we can 

mitigate the situation if necessary. And this isn’t necessarily building enhancement structures or replacing 

culverts. New information can be used to justify policies and management, such as a reduction in salmon 

harvests or other conservation measures. The main point is that the complexity of climate change is 

bringing up lots of questions, and managers would do well to establish specific needs and research 

priorities before getting started. 

CRITIQUE 

General Approach 

The initial steps that were suggested for this assessment follow a rational and logical progression – 

defining the assessment area, identifying the resource values, describing the sensitivity of these values, 

identifying stressors, and determining exposure. Identifying the resource values is especially important 

because it focuses the analysis on the relevant issues. 

The other Forests compared all of their watersheds to determine which were the most vulnerable but this 

was not a priority for the Chugach. As mentioned earlier, most of the watersheds have little or no 

development – 99% of the Forest is in roadless areas. Although climate change can affect resources in all 

of the watersheds, I felt that it was unlikely that managers would conduct mitigation measures in pristine 

areas. 

Not ranking the relative vulnerability of the watersheds may be one weakness of this assessment. The 

assessment does not show, for example, that the fisheries values of the Kenai River system (with 

headwaters on National Forest land) far outweigh the Resurrection Creek fisheries. However, Chugach 

managers don’t have more than a half dozen developed watersheds to look at, so they have the luxury of 

being able to look closely at each watershed. Given the low levels of development in the Kenai area and 

knowing that the climate change conditions will be similar, managers will still need to be working on a 

site-specific scale, watershed by watershed, to develop meaningful plans and establish project priorities. 

Data Availability 

There is a good deal of climate change information available from the UAF SNAP program, from raw 

GCM data to ready-made maps and graphs. Other websites have historic evapotranspiration estimates and 

other parameters that could be useful in more extensive analyses. 

Predicting change for streamflow and runoff timing in coastal Alaska is difficult due to several conflicting 

factors. Climate change models predict warmer temperatures and increased precipitation for coastal 

Alaska, but given the high elevations of the area, reductions in snowpack at lower elevations may be 

offset by higher precipitation and more snow at higher elevations. Earlier melting of the snowpack may 

be compensated for by increased glacial melting augmenting flows in late summer, – at least until the 

glaciers are gone. Most of the literature agreed that glaciers were melting more rapidly, but increased 

snowpacks in coastal Alaskan mountains was only mentioned as a possibility. 



As a result, the main data gap was an estimate of the future change in streamflows, snowpacks, runoff 

timing, and other parameters. My assumption is that this information is available from VIC and other 

models for the lower 48 states, but I am not aware that such data are available for Alaska yet. The limited 

numbers of stream-gauging stations, limited duration of station operation, and the limited number of 

weather data sites in remote areas may be part of the reason. In any case, the data did not appear to be 

readily available, so I turned my focus to qualitative assessments. 

Other data gaps included long-term water temperature data and stream height/flood level data. Having 

more specific data would have added more certainty to some statements and conclusions, but overall I 

think the general concepts are valid. 

The accuracy of the data provided by the models appeared to be a little questionable at times. For some 

areas near Cordova, the maps don’t always fit the topography, which may reflect the extrapolations 

between distant weather stations or distance from the ocean. The 2 km cells may also add some 

uncertainty if one is trying to analyze a relatively small area. However, if one is only looking for trends, 

small discrepancies may not be a concern. 

The variation among models also raises some questions. The SNAP website states that the variability 

among the models is generally in the range of 0-4 °F and 0-0.7 inches for precipitation. Four degrees is a 

large range when one is looking at winter temperatures that are near freezing. For Hope, where conditions 

are relatively dry, the range of variability for precipitation is often greater. There is also the question of 

whether an average of five models is any more accurate than any single model. Thus, if one were to do a 

quantitative analysis, there may be problems. However, the models all agree in the general trends, which 

should be sufficient for some types of analysis. 

Assessing Risk 

One of the suggested methods for assessing overall watershed vulnerability was to create a risk matrix, 

comparing various attributes such as road density or slope, values such as fish populations, predicted 

climate change parameters, and then assign risk levels on a low to high scale. The total scores would be 

used to determine the most vulnerable watersheds. This process did not appear to be applicable for the 

Chugach National Forest, where most of the watersheds are undisturbed, road densities are uniformly low 

or zero, and the risks to fish and other wildlife from the predicted climate changes are unclear. 

Assigning different levels of risk seemed to be subjective, given the wide differences between the 

ecosystems. While winter temperatures are expected to increase by about 3.7 °C for both Hope and 

Cordova, the effect in Cordova will be much greater since low-elevation winter temperatures are hovering 

around the freezing point. Similarly, larger precipitation increases in Cordova are probably less 

meaningful, given the currently high precipitation. Also, some watersheds may have greater fire hazards, 

while others may have more valuable fish, so the comparisons may not be equal. 

With the limited number of developed watersheds, it didn’t seem necessary to rank them to determine 

which are the most vulnerable. For the Chugach, it seems simpler to identify the specific issues for each 

watershed on its own, since there are only a few to analyze. 

The other problem is determining the magnitude of adverse effects from climate change over existing 

conditions. As discussed, the predicted increases in temperature and precipitation are well within the 

historical variability, although more extreme weather events are expected. While one can intuitively say 

that greater precipitation could lead to greater erosion and landslides, it may be difficult to argue that 

another 6 inches of rain will increase landslides in a watershed that already receives a mean of 177 inches. 



To really answer some of these questions, it will take a good deal of professional knowledge and 

modeling expertise to predict the effects with more certainty. 

The ability to assess risk is also difficult when the biological effects are unknown. Certainly there is the 

potential for major disruptions to the food chains, salmon life histories, and aquatic invertebrate life 

cycles due to increased water temperatures. The absolute temperature probably isn’t the biggest factor, 

but simply that water temperatures will change for species adapted to the former conditions. Thus, all 

watersheds may have similar disruptions. The question of risk then becomes whether the organisms can or 

cannot easily adapt to these new conditions, and that is unknown. 

Implementation 

Before conducting a vulnerability assessment, managers need to be able to commit a good deal of time 

and have knowledgeable personnel with the appropriate technical skills. For a team with no previous 

climate change experience, a large amount of time can be spent learning about the data that are available 

and reviewing the literature. Specialists from all fields will be needed to identify values and determine 

effects. A diverse, interdisciplinary group will also know more about existing plans, strategies, and what 

actions are really possible. Thus, a large commitment of time and personnel is required to do the 

assessment, and even more to turn the findings into a plan of action. 

It may be better for the Forest Service to establish an Enterprise Team that has expertise using climate 

change data and models. A large part of the learning curve can be eliminated in this fashion. Local 

specialists will still be needed to identify site-specific values and issues. The team could also develop a 

stock set of mitigation prescriptions for a variety of circumstances. 

REFERENCES 

Arendt, A.A, Echelmeyer, K.A., Harrison, W.D., Lingle, C.S., and V. B. Valentine. 2010. Rapid 

Wastage of Alaska Glaciers and Their Contribution to Rising Sea Level. Science 297:382-386. 

Bair, B. P. Powers, and A. Olegario. 2002. Resurrection Creek stream channel and riparian restoration 

analysis, river kilometer 8.0-9.3. Project Report for the USDA Forest Service by the Wind River 

Watershed Restoration Team. 

Bakke, P. 2008. Physical processes and climate change: A guide for biologists. Unpublished report. U.S. 

Fish and Wildlife Service. Available: http://www.stream.fs.fed.us/ 

publications/documentsNotStream.html. – states depositional areas most sensitive to change. Instream 

structures need to be more robust, redundant in areas where channel change more likely, or better, passive 

means such as wider riparian buffers. 

Blanchet, D. 1983. Evaluation of recent channel changes on the Scott River near Cordova, Alaska. 

USDA Forest Service, Chugach National Forest, Anchorage, AK. 

Boggild, Carl E., Niels Reeh, and Hans Oerter. 1994. Modeling ablation and mass-balance sensitivity 

to climate change of Stormstrmmen, Northeast Greenland. Global and Planetary Change 9:79-90. 

Botz, J., G. Hollowell, J. Bell, R. Brenner, and S. Moffitt. 2010. Fishery Management Report No. 10- 

55. 2009 Prince William Sound area finfish management report. Alaska Department of Fish and Game, 

Division of Commercial Fisheries, Cordova. 



Bryant, M. D. 2009. Global climate change and potential effects on Pacific salmonids in freshwater 

ecosystems of southeast Alaska. Climatic Change 95:169-193. 

Chittenden, C. M., Beamish, R. J., and R. S. McKinley. 2009. A critical review of Pacific salmon 

marine research relating to climate. – ICES Journal of Marine Science, 66: 2195–2204. 

Copper River Watershed Project. 2011. Prioritizing Fish Passage Improvement Projects in the Copper 

River Watershed. Cordova, AK. 

Copper River Watershed Project and Prince William Sound/ Copper River Marketing Association. 

2011. Personal communication. Kate Alexander (CRWP) sent an e-mail to numerous agencies, 

organizations, Native groups, fishing groups, and others, soliciting interest in a fisheries sustainability 

plan for the Prince William Sound and Copper River fishing areas. 

Crawford, R.E. 2010. Hydroacoustic visualization of Eyak Lake’s bathymetric features: 

Updating the AMSA bathymetric map of Eyak Lake. Prepared for the Copper River Watershed Project. 

Cordova, AK. 

Criscitiello, A., M. A. Kelly, and B. Tremblay. 2010. The response of Taku and Lemon Creek glaciers 

to climate. Arctic, Antarctic, and Alpine Research 42(1):34-44. 

Dowdeswell, J.A. et al. 1997. The Mass Balance of Circum-Arctic Glaciers and Recent Climate Change. 

Quaternary Research 48, 1–14. States that no uniform trend for Arctic, but high loss in mass balance in 

Alaska due to higher summer temperatures, increases for maritime Scandinavia and Iceland due to 

increased precipitation. 

Ecology and Environment Incorporated. 2006. Community Wildfire Protection Plan for At-Risk 

Communities Near Chugach National Forest, Alaska: Hope, Sunrise, Summit Lake. Prepared in 

cooperation with the USDA Forest Service, Chugach National Forest. 

http://www2.borough.kenai.ak.us/SBB/documents/CWPP/Hope-Sunrise- 

Summit%20CWPP%20Final%20Draft%2007-28-06.pdf 

Farr, W.A. and A.S. Harris. 1979. Site index of Sitka spruce along the Pacific coast related to latitude 

and temperatures. Forest Science 25:145-153. 

Furniss, M.J. and 12 others. 2010. Water, climate change, and forests: watershed stewardship for a 

changing climate. General Technical Report PNW-GTR-812. Portland, OR: U.S. Department of 

Agriculture, Forest Service, Pacific Northwest Research Station. 

Hamlet, A. F., and D. P. Lettenmaier. 2007. Effects of 20th century warming and climate variability on 

flood risk in the western U.S., Water Resources. Research, 43, W06427. 

Haufler, J.B., C.A. Mehl, and S. Yeats. 2010. Climate change: anticipated effects on ecosystem services 

and potential actions by the Alaska Region, U.S. Forest Service. Ecosystem Management Research 

Institute, Seeley Lake, Montana, USA. 

Hilborn, R., T.P. Quinn, D. Schindler, and D. Rogers. 2003. Biocomplexity and fisheries 

sustainability. Proceedings of the National Academy of Science 100:11:6564-6568. 

Hitch, Kenneth E. 1995. Letter from the U.S. Army Corps of Engineers to Scott Janke, City Manager, 

City of Cordova. Available at the Chugach National Forest, Cordova Ranger District. 



Hodges, K. 2000. Project status of proposed Scott River dike. Unpublished Forest Service report. 

Cordova Ranger District, Cordova, AK. 

Hodges, K., S. Greenwood, and K. Buckley. 1995. Changes in cutthroat trout spawning habitat in the 

Eyak Lake watershed, Cordova, Alaska. Unpublished Forest Service report available at the Chugach 

National Forest, Cordova Ranger District. 

ISAB. 2007. Independent Science Advisory Board for the Northwest Power and Conservation Council, 

Columbia River Basin Indian Tribes, and National Marine Fisheries Service. Climate Change Impacts on 

Columbia River Basin Fish and Wildlife. ISAB Climate Change Report 2007-2. 

Kalli, G. and D. Blanchet. 2001. Resurrection Creek Watershed Association Hydrologic Assessment. 

USDA Forest Service, Chugach National Forest, unpublished internal report. 

Kenai Peninsula Borough. 2004. Interagency all lands/all hands action plan for fire prevention and 

protection, hazardous fuel reduction, forest health and ecosystem restoration, community assistance in 

Alaska’s Kenai Peninsula Borough. 

Kenai Peninsula Borough. 2011. All-Hazard Mitigation Plan. Website 

http://www2.borough.kenai.ak.us/emergency/HazMit/plan.htm 

Lang, D. W. 2010. A survey of sport fish use on the Copper River Delta, Alaska. General Technical 

Report PNW-GTR-814. U.S. Department of Agriculture, Forest Service, Pacific Northwest Research 

Station. 

Lang, D.W., G.H. Reeves, J.D. Hall, and M.S. Wipfli. 2006. The influence of fall-spawning coho 

salmon (Oncorhynchus kisutch) on growth and production of juvenile coho salmon rearing in beaver 

ponds on the Copper River Delta, Alaska. Canadian Journal of Fisheries and Aquatic Sciences. 63:917- 

930. 

Martin, A.E., M.S. Wipfli, and R.E. Spangler. 2010. Aquatic community responses to salmon carcass 

analog and wood bundle additions in restored floodplain habitats in an Alaskan stream. Transactions of 

the American Fisheries Society 139:1828-1845. 

McClure, C.J.W., B.W. Bolek, K. McDonald, and G.E. Hill. 2012. Climate change and the decline of a 

once common bird. Ecology and Evolution. 2(2):370-378. 

McCullough, D. 1999. A Review and Synthesis of Effects of Alterations to the Water Temperature 

Regime on Freshwater Life Stages of Salmonids, with Special Reference to Chinook Salmon. Columbia 

Intertribal Fisheries Commission, Portland, OR. Prepared for the U.S. Environmental Protection Agency 

Region 10. EPA 910-R-99-010. 

Montgomery, D., J. Buffington, N.P. Peterson, D. Schuett-Hames, and T.P. Quinn. 1996. Stream-bed 

scour, egg burial depths, and the influence of salmonid spawning on bed surface mobility and embryo 

survival. Canadian Journal of Fisheries and Aquatic Sciences 53:1061-1070. 

Montgomery, D., E. Beamer, G. Pess, and T.P. Quinn. 1999. Channel type and salmonid spawning 

distribution and abundance. Canadian Journal of Fisheries and Aquatic Sciences. 56: 377–387. 



National Aeronautics and Space Administration. Global Land Data Assimilation System 

http://gdata1.sci.gsfc.nasa.gov/daac-bin/G3/gui.cgi?instance_id=GLDAS10_M viewed 10/2011. 

National Oceanic and Atmospheric Administration, National Marine Fisheries Service. 1996. 

Coastal salmon restoration: working guidance for comprehensive salmon restoration initiatives on the 

Pacific coast. 

Piper, R. I. McElwain, L. Orme, J. McCraren, L. Fowler, and J. Leonard. 1982. Fish hatchery 

management, third printing. USDI Fish and Wildlife Service. Washington, D.C. 

Poff, L. N., M. Brinson, and J. Day. 2002. Aquatic ecosystems and global climate change. Potential 

impacts on inland freshwater and coastal wetland ecosystems in the United States. Report prepared for 

the Pew Center on Climate Change. 

Potkin, M. 1997. Fire history disturbance of the Kenai Peninsula mountainous portion of the Chugach 

National Forest. Unpublished USDA Forest Service report. Available online at 

http://depts.washington.edu/nwfire/publication/Potkin_1997.pdf 

Professional Fishery Consultants. 1985. Eyak Lake AMSA Cooperative Management Plan. Prepared 

for the Eyak Lake AMSA (area meriting special attention) Study Team. Funding from the Alaska Coastal 

Management Program, Office of Ocean and Coastal Resource Management; and US Department of 

Commerce, National Oceanic and Atmospheric Administration. 

Ramstad, K.M., C.J. Foote, J.B. Olsen, and D. Rogers. 2005. Genetic and phenotypic evidence of 

reproductive isolation between seasonal runs of sockeye salmon in Bear Lake, Alaska. Transactions of the 

American Fisheries Society 132: 997-1013. 

Reeves, G. et al. 2003. Aquatic and Riparian Effectiveness Monitoring Plan for the Northwest Forest 

Plan. Gen. Tech. Rep. PNW-GTR-577. Portland, OR: U.S. Department of 

Agriculture, Forest Service, Pacific Northwest Research Station. 70 p. 

Reynolds, K. M. 2001. Fuzzy logic knowledge bases in integrated landscape assessment: examples and 

possibilities. Gen. Tech. Rep. PNW- GTR-521. Portland, OR: U.S. Department of Agriculture, Forest 

Service, Pacific Northwest Research Station. 24 p. 

Rothwell, E. and A. Bidlack. 2011. Using hydrology equations to estimate a water budget and data gaps, 

Eyak Lake, AK. Poster presentation at the Copper River Delta Symposium, March 22-24, 2011, 

Cordova, AK. 

Ruff, C.P. et al. 2011. Temperature-associated population diversity in salmon confers benefits to mobile 

consumers. Ecology 92(11):2073-2084. 

Scenarios Network for Alaska Planning. 2011. Community charts. Available on the internet at: 

http://www.snap.uaf.edu/home. 

Sea-Run Fisheries. 2006. Fish, wildlife and water quality monitoring for 2005 per FERC license article 

410 Power Creek Hydroelectric Project FERC project no. 11243 Cordova, Alaska. For Cordova Electric 

Cooperative, Cordova, AK. 

Seppälä, R., A. Buck, and P. Katila. (eds.). 2009. Adaptation of Forests and People to Climate Change. 

A Global Assessment Report. IUFRO World Series Volume 22. Helsinki. 224 p. 



Sillman, J. and E. Roeckner. 2008. Indices for extreme events in projections of anthropogenic 

climate change. Climatic Change 86:83–104. 

Tonina, D. and J. McKean. 2010. Climate change impact on salmonid spawning in low-gradient streams 

in central Idaho, USA. 9th International Conference on Hydroinformatics HIC 2010, Tianjin, China. 

Viewed online October 2011. 

http://www.fs.fed.us/rm/boise/AWAE/scientists/profiles/mckean/rmrs_2011_toninad002.pdf 

University of Alaska, Fairbanks. 2011. Scenarios Network for Alaska Planning. Community charts. 

Available on the internet at: http://www.snap.uaf.edu/charts.php. 

U.S. Census Bureau. http://factfinder.census.gov/servlet/ACSSAFFFacts 

USDA Forest Service, Alaska Region, Chugach National Forest. 2002. Revised land and resource 

management plan. Final environmental impact statement. R10-MB-480d. 

Winder, M. and D. Schindler. 2004. Climate change uncouples trophic interactions in an aquatic 

ecosystem. Ecology, 85(8):2100–2106. 


The Watershed Vulnerability Assessment Guild in Salt Lake City, September 2010. 

From left to right: Christine Mai (Shasta-Trinity NF), Dave Cleaves (Climate Change Advisor to the Chief), 

Dana Kuntzsch (Chugach NF), John Chatel (Sawtooth NF), Carol Howe (GMUG NF), Dan Cenderelli 

(Stream Systems Technology Center), Laura Jungst (Helena NF), Mark Weinhold (White River NF), Polly 

Hays (Rocky Mountain Region), Dale Higgins (Chequamegon-Nicolet NF), Joan Louie (Gallatin NF, 

currently Northern Region), Alan Clingenpeel (Ouachita NF – retired), Michael Furniss (Pacific Northwest 

Research Station), Ken Roby (Lassen NF – retired), Caty Clifton (Umatilla NF), Jamey Lowdermilk (Guest, 

Helena NF), Scott Vuono (Guest, Sawtooth NF), Karen Bennett (Pacific Northwest Region), Brian Staab 

(Pacific Northwest Region), Ann Carlson (Lassen NF), Kerry Overton (Rocky Mountain Research Station), 

Ken Hodges (Chugach NF). Not pictured: Ralph Martinez (Plumas National Forest) and Rory Steinke 

(Coconino National Forest).

watervulnerability

You also want an ePaper? Increase the reach of your titles

Delete template?

Save as template?